Medicine A path through The Default Library ======================================== How the body works, how it fails, how doctors reason, and what medicine can and can't do. Step 1. Biochemistry: The Chemistry That Keeps Itself Going ----------------------------------------------------------- Why: The molecules and metabolism behind every disease. A chain of amino acids folds itself, an enzyme speeds a reaction a million-fold, and a cell runs on a rechargeable molecule it turns over every minute. Biochemistry is the course where the molecules of life are treated as chemistry, with a ledger. .chips button{min-width:0} .row2{display:flex;gap:8px;flex-wrap:wrap} .row2 button{flex:1 1 90px;min-width:0} Eat a piece of bread and, a few hours later, its glucose is a trickle of electrons moving along proteins inside your mitochondria. A single genetic typo in a haemoglobin gene turns one amino acid into another, and red cells bend into sickles. A medicine blocks one enzyme and a symptom eases. In each case the story can be told in molecules, in numbers, and in energy bookkeeping. Biochemistry asks what those molecules are, how they fold and catalyse, and how a cell pays for everything. It is the bridge between the chemistry of carbon and the biology of cells, and it sits on both routes: chemists take it to see where their subject leads, and biologists take it to see what their subject is made of. The course in one box Who takes itChemistry majors, usually in year 3 or 4; biology, biomedical and nutrition majors, usually in year 2 or 3; and, as a prerequisite, most students heading for medicine, pharmacy or dentistry. Programmes shaped by the American Chemical Society guidelines list it among the foundations of a chemistry degree, and the Vision and Change framework puts "structure and function", "energy and matter" and "information flow" at the centre of undergraduate biology, which is exactly its territory. What it is forExplaining how the large molecules of life are built, how they speed up and steer reactions, and how cells store and spend energy. Comes afterOrganic Chemistry: Structure & Mechanism (35/38). Leads toMolecular Biology (35/45), then Physiology (35/49) and Microbiology (35/50), plus Cell Biology (35/44) as its companion. It is also the chemistry under Genetics (35/46) and Plant Biology (35/53). What this piece gives youYou will not be able to balance a metabolic pathway from memory. You will see why shape is everything for a protein, what Michaelis–Menten kinetics predict, how a cell can pull energy from sugar in steps, why fat stores more energy than starch, and what AlphaFold did and did not solve. Easier first reads: Fermentation: Letting Microbes Do the Cooking (14/08), The Science of Bread (14/07), and Entropy: Why You Can't Unstir Your Coffee (13/12) for the idea of free energy. This describes how science works; it is not medical advice. Twenty amino acids, one folded shape Proteins are chains of amino acids, twenty kinds in the standard genetic code, joined by peptide bonds. Each kind has a different side chain: some are oily and avoid water (leucine, valine), some carry a charge (lysine is positive, aspartate negative), some can form hydrogen bonds (serine). A chain of a few hundred of these is a protein. The sequence is written in DNA, but what does the work is the shape: oily side chains bury themselves in the core, charged ones face the water, and the chain settles into helices and sheets that pack into a compact fold. In 1961 Christian Anfinsen showed that the enzyme ribonuclease, unfolded with chemicals, refolded by itself into an active form when the chemicals were removed. The information for the shape is therefore in the sequence. He shared the 1972 Nobel Prize in Chemistry (with Stanford Moore and William Stein). It has caveats: some proteins need helper "chaperones", and some work while floppy. But it set the problem. The sharp question is Levinthal's. A chain cannot find its shape by trying them all. Cyrus Levinthal estimated in 1969 on the order of 10300 possible shapes for a typical protein, yet real ones fold in milliseconds to seconds. The slider counts it your way: if each of n residues has just 3 positions, there are 3n shapes, and at a trillion tries a second the search for a 100-residue chain would last about 1027 years. Proteins do not search; they slide downhill on an energy landscape shaped like a funnel. Interactive · Why a protein cannot search Residues in the chain Shapes per residue A thought experiment, not a model of folding: kn shapes, one tried every 10−13 s (a bond vibration), against the universe's age of 13.8 billion years. Real proteins fold in about 10−3 to 10 s. Log scale. A single change can matter enormously. In sickle-cell disease, glutamic acid at position 6 of the β-globin chain of haemoglobin is replaced by valine. That oily side chain on the surface makes haemoglobin molecules stick into fibres when they let go of oxygen, and bend the red cell. Carriers of one copy are partly protected against severe malaria, which is why the variant is common where malaria has been. Genes and environment are in conversation here, a theme of Genetics (35/46). Enzymes: catalysts that set the pace of life Most reactions that a cell needs would take years at body temperature. Enzymes make them take milliseconds. An enzyme does not change where a reaction ends up (the equilibrium); it lowers the barrier on the way, by holding the substrate in a pocket shaped to stabilise the transition state. Carbonic anhydrase, which interconverts carbon dioxide and bicarbonate in your blood, can turn over about a million substrate molecules per second per enzyme, among the fastest known. In 1913 Leonor Michaelis and Maud Menten wrote down the simplest model that explains how rate depends on substrate concentration. At low [S] the enzyme is mostly empty, so the rate rises in proportion. At high [S] every enzyme is busy and the rate saturates at Vmax. The Km is the substrate concentration at which the rate is half of Vmax: a small Km means the enzyme grabs its substrate even when it is scarce. v = Vmax · [S] / (Km + [S])rate rises with substrate, then saturates; Km is the [S] where v = Vmax/2 A competitive inhibitor resembles the substrate and occupies the pocket. It does not change Vmax (flood the system with substrate and you win), but it raises the apparent Km by a factor (1 + [I]/Ki). Many drugs are inhibitors of enzymes: ibuprofen reversibly blocks the cyclooxygenase enzymes, and aspirin does it by chemically modifying them. Try the sliders. Interactive · Michaelis–Menten, with an inhibitor Substrate [S] Km Competitive inhibitor, [I] / Ki Illustrative units: Vmax = 100 µM/s. Solid: with the inhibitor; dashed: without. Curves computed from v = Vmax[S]/(Km,app + [S]). Notice what the readout says: at [S] = Km the rate is exactly half of maximum, and at 10 × Km it is already 91%, so doubling the substrate again barely helps. That saturation is why a cell regulates enzymes by controlling the enzyme, not the substrate. ATP: the energy ledger's currency A cell cannot burn sugar like a flame; the heat would cook it. Instead it moves energy in small, controlled steps, and the common change is ATP (adenosine triphosphate). Splitting its last phosphate off to give ADP and phosphate releases energy that other reactions can borrow when they are coupled to it. Under standard biochemical conditions (1 M of everything, pH 7) the free energy change is −30.5 kJ per mole. In a living cell the concentrations are far from standard, and the free energy depends on them: ΔG = ΔG°′ + RT ln([ADP][Pi] / [ATP])free energy of ATP hydrolysis; R = 8.314 J/(mol·K), T = 310 K at body temperature, so RT ≈ 2.58 kJ/mol Textbooks quote about −50 to −65 kJ/mol inside cells. The reason is striking: at equilibrium ATP would be almost entirely broken down, and a cell keeps its ATP/ADP ratio many orders of magnitude above that. It is held there by constant work, and a human turns over roughly their own body weight in ATP in a day (an estimate), recycling it between ATP and ADP, not accumulating it. Interactive · How far from equilibrium? Resting cell Exhausted [ATP] [ADP] [Pi] Illustrative concentrations in mM, 37 °C, ΔG°′ = −30.5 kJ/mol. Real cells vary by tissue and state, and free Mg2+ matters; this is the textbook estimate. Metabolism as bookkeeping: glucose to ATP Metabolism is a set of linked reactions, and the neat way to hold it in your head is a ledger with three columns: ATP, and two electron carriers, NADH and FADH2, which carry electrons to the cell's power plant. Glycolysis, in the cytosol, splits one glucose into two pyruvate, and pays a net 2 ATP and 2 NADH. With oxygen, pyruvate goes into the mitochondrion, giving 2 more NADH (as it becomes acetyl-CoA), and the citric acid cycle (Hans Krebs, 1937) turns the rest into carbon dioxide, with 6 NADH, 2 FADH2 and 2 more ATP-equivalents. All that is the "earnings"; the real payout comes when the carriers deposit their electrons into the respiratory chain, which pumps protons across the inner membrane. Peter Mitchell proposed in 1961 that the resulting proton gradient, not a chemical intermediate, drives ATP synthase, a tiny rotary motor; his chemiosmotic theory won the 1978 Nobel Prize in Chemistry. The old textbooks said one glucose gives 36 to 38 ATP. Newer measurements put it at about 30 to 32, because a proton-driven motor does not pay in whole molecules. The ledger lets you see why the range exists. Interactive · The ATP ledger, step by step Next stage Reset Malate shuttle Glycerol-phosphate shuttle Modern (2.5 and 1.5) Old textbook (3 and 2) Per glucose. The ATP yield per NADH (about 2.5) and FADH2 (about 1.5) are modern estimates; older textbooks used 3 and 2. Cytosolic NADH enters through a shuttle, which costs a little in some tissues. All numbers are textbook estimates. Without oxygen, the cell stops after glycolysis and must recycle its NADH by making lactate or, in yeast, ethanol: two ATP per glucose instead of about thirty. That is the arithmetic behind a sprinter's burning legs and a baker's rising dough, and it is why oxygen matters. Lipids, sugars and membranes: fuel, walls and labels Three more families fill out the picture. FamilyBuilt fromRoles Carbohydratessugars (glucose, fructose) joined in chainsquick fuel (starch, glycogen); structure (cellulose); labels on cell surfaces Lipidsfatty acids, glycerol, cholesteroldense energy store; membranes; signals (steroid hormones) Proteinsamino acidscatalysis, structure, transport, signalling The small difference in how sugar units are joined has large effects: starch and cellulose are both glucose polymers, but the link is shaped differently in each, so humans digest starch and cannot digest cellulose. Fats carry about 37 kJ per gram (9 kcal) against about 17 kJ (4 kcal) for carbohydrate and protein, because a fat is more reduced, with more electrons to hand to the respiratory chain. That is why animals store energy as fat. Membranes come from a simpler fact: a phospholipid has a water-loving head and two oily tails. In water such molecules assemble themselves into a double layer, tails inward, about 5 nanometres thick, which is the same physics that lets soap form droplets around grease. Proteins embedded in that layer do the transport and sensing; see Cell Biology (35/44). Nucleic acids as molecules DNA and RNA are chains of nucleotides: a sugar, a phosphate and a base. The backbone is sugar–phosphate; the bases (A, C, G, T in DNA; U replaces T in RNA) stick out. In 1953 James Watson and Francis Crick proposed the double helix, using X-ray data from Rosalind Franklin's group and Maurice Wilkins's; the 1962 Nobel Prize went to Watson, Crick and Wilkins. The key chemistry is complementary pairing: A binds T with two hydrogen bonds, and G binds C with three. Erwin Chargaff had found the matching counts of A with T, and G with C, years earlier. How the code is read is the job of Molecular Biology (35/45); here the point is simpler: a G–C pair holds more tightly than an A–T pair. For a short DNA piece, a rule of thumb (the Wallace rule) puts the melting temperature at about 2 °C for each A or T and 4 °C for each G or C. Longer pieces and real salt conditions need more careful models, which is why labs use calculators. Interactive · How tightly does a short duplex hold? Length G–C share Wallace rule, valid only for short pieces (about 14 bases or fewer): Tm ≈ 2(A+T) + 4(G+C), in °C. Dark = G–C (three hydrogen bonds), light = A–T (two). AlphaFold and the protein-structure problem Knowing a protein's sequence is not the same as knowing its shape. For decades the only way to get a shape was experiment: X-ray crystallography, NMR, and more recently cryo-electron microscopy, each a long project. The Protein Data Bank, started in 1971, now holds well over 200,000 experimental structures, but known protein sequences number in the hundreds of millions. Predicting the shape from the sequence became a competition, CASP, held every two years. In November 2020 DeepMind's AlphaFold 2 reached accuracy in CASP14 that was comparable to experimental structures for many targets, a result the organisers called a breakthrough. The method was published in 2021 (Jumper and colleagues) and in 2022 predictions for about 200 million proteins were released in a public database. The 2024 Nobel Prize in Chemistry went half to David Baker for computational protein design and half to Demis Hassabis and John Jumper for structure prediction. What it did not do is end the subject. A prediction is a model with a confidence score, not a measurement. It does not by itself explain how a chain folds (Levinthal's question), how a protein moves or how it behaves when it binds a drug or another protein, and it is less certain for floppy regions and rare folds. Experimental structures still check it, and they train it. Treat it as a very good map, drawn from the territory that experiments surveyed. What a graduate just sees A protein's job is set by its shape, and its shape is set by its sequence and by water. One amino acid out of hundreds can change everything, and effects depend on environment. Enzymes lower barriers and never move the equilibrium; rates saturate at Vmax. Cells keep ATP enormously far from equilibrium, and that distance is the energy they spend. Metabolism is a ledger: sugar becomes carriers (NADH, FADH2), carriers become a proton gradient, the gradient becomes ATP. Fat is dense fuel because it is more reduced than sugar. Membranes assemble themselves from molecules with a wet head and an oily tail. G–C pairs are stronger than A–T pairs: three hydrogen bonds against two. A predicted structure is a model with a confidence level, not a measurement. Where it shows up next Molecular Biology (35/45): how DNA is copied and read, and how cells control which genes are on. Physiology (35/49): hormones, muscle and the fuel switches between fed and fasting states. Microbiology (35/50): bacteria differ chiefly in which metabolism they run, with or without oxygen. Plant Biology (35/53): photosynthesis is respiration in reverse, with light paying the bill. In the world: drug design, enzyme-based detergents and diagnostics, brewing and baking, sports science and diabetes care. If you want to go deeper 📘 Textbook. David Nelson and Michael Cox, Lehninger Principles of Biochemistry (Macmillan / W. H. Freeman, 8th edition). The standard in many courses; strong on energetics. 📘 Textbook. Jeremy Berg, John Tymoczko, Gregory Gatto and Lubert Stryer, Biochemistry (Macmillan / W. H. Freeman, 9th edition). Known for clear diagrams and for connecting to medicine. 📘 Gentler. Nick Lane, Power, Sex, Suicide: Mitochondria and the Meaning of Life (Oxford University Press, 2005; 2nd ed. 2018). A popular account of energy in cells. 🌐 Free course. MIT OpenCourseWare 7.05 General Biochemistry. Also the free OpenStax Biology 2e. The takeaway Life's molecules obey the same chemistry as everything else; what is special is how they are arranged. Shape follows from sequence, enzymes set the pace, ATP is held far from equilibrium and spent in small steps, and a good ledger tells you where the energy of a sugar goes. Try this: chew an unsweetened cracker for a full minute and notice when it starts to taste sweet. An enzyme in your saliva is snipping starch into sugars, and you are tasting the product. Sources & Further Reading Sources. Anfinsen (1961), Nobel Prize in Chemistry 1972 (with Moore and Stein). Levinthal (1969), "How to fold graciously": estimate of about 10300 conformations. Sickle-cell: Glu6Val in β-globin; carriers' partial protection against malaria (established in the 1950s). Carbonic anhydrase turnover about 106 per second: Lindskog, Pharmacology & Therapeutics (1997). Michaelis and Menten (1913). ATP hydrolysis ΔG°′ = −30.5 kJ/mol and cellular values near −50 to −65 kJ/mol: Nelson and Cox, Lehninger. Krebs (1937); Mitchell, chemiosmotic hypothesis (1961), Nobel Prize in Chemistry 1978. ATP yield of about 30 to 32 per glucose: Lehninger and other recent textbooks (older editions 36 to 38). Fat and carbohydrate energy: Atwater factors (9 and 4 kcal/g). Watson and Crick (1953); Nobel Prize in Physiology or Medicine 1962; Chargaff's rules; Wallace rule for short oligonucleotides. Protein Data Bank (announced 1971, Nature New Biology). AlphaFold: CASP14 (November 2020); Jumper et al., Nature 596 (2021); AlphaFold Database release of about 200 million structures (July 2022); Nobel Prize in Chemistry 2024 (Baker; Hassabis and Jumper). Numbers in the interactives are illustrative or textbook estimates and are labelled. This is not medical advice. 🌐 MIT OCW 7.05 General Biochemistry 🌐 OpenStax Biology 2e 📘 Nelson and Cox, Lehninger Principles of Biochemistry 📘 Lane, Power, Sex, Suicide Step 2. Cell Biology: The Smallest Unit That Is Alive ----------------------------------------------------- Why: Cells, and cancer as their breakdown. Every living thing from a bacterium to a blue whale is made of cells, and no cell is very large. The course shows why, what is inside them, how they talk, how they divide, and how that division goes wrong in cancer. .chips button{min-width:0} .row2{display:flex;gap:8px;flex-wrap:wrap} .row2 button{flex:1 1 90px;min-width:0} A drop of pond water under a simple microscope teems with things that were not visible a moment ago: one swims, one divides, one eats another. In 1665 Robert Hooke looked at thin slices of cork and drew little chambers he called "cells", and a few years later Antonie van Leeuwenhoek saw living ones swim. By 1838 and 1839 Matthias Schleiden and Theodor Schwann had proposed that all plants and animals are built of cells, and cell theory was born. Later, in 1855, Rudolf Virchow popularised the idea that new cells come only from existing ones (the phrase is often credited to him, and Robert Remak's earlier work is part of that story). Cell biology is the course that takes that idea seriously. Whatever a living thing does, a cell does it: it keeps an inside different from the outside, it turns food into work, it makes copies of itself, and it listens to its neighbours. A graduate keeps a mental picture of that small, crowded, busy room. The course in one box Who takes itEvery biology major, usually in year 1 or 2; biochemistry, neuroscience, nursing and medical students too. Programmes shaped by Vision and Change (the AAAS and NSF framework for undergraduate biology), MIT's Course 7 and Cambridge's Natural Sciences all place the cell at the base of biology, early. What it is forSeeing life as a collection of small, self-maintaining units, and learning what is inside each. Comes afterGeneral Chemistry (35/36). Leads toMolecular Biology (35/45), then Physiology (35/49), Plant Biology (35/53) and, through those, Genetics (35/46) and Microbiology (35/50). Biochemistry (35/43) is its chemical companion. What this piece gives youYou will not be able to draw every organelle from memory. You will see why cells are as small as they are, what a membrane's voltage means, how transport inside a cell is solved, why cancer is a disease of the cell cycle, and what a microscope can and cannot show. Easier first reads: Your Immune System, Day by Day (01/10) and Biological Psychology (35/20), where the neuron's voltage is a cell-biology story. This describes how biology works and is not medical advice. Why cells are small Cells span a huge range, from a bacterium about a micrometre long (a thousandth of a millimetre) to a human cell near ten, to an egg that is visible to the naked eye. Yet there is a ceiling, and the reason is geometry. A cell takes in food and oxygen through its surface, but burns it throughout its volume. Double a cell's radius and its surface grows 4 times, its volume 8 times: the supply line grows more slowly than the demand. For a sphere, surface-to-volume ratio is 3 divided by the radius, so smaller is better. The second limit is time. Inside a cell, many molecules move by diffusion, a random walk, and the time it takes to cross a distance grows with the square of the distance. A small molecule diffusing at about 100 µm² per second (an illustrative value for the cytoplasm) crosses a 1 µm bacterium in about a thousandth of a second, a 10 µm cell in about a sixth of a second, and a giant cell 1 mm in radius in around half an hour. t ≈ r² / (6D), and surface-to-volume = 3 / rfor a sphere of radius r; D is the diffusion coefficient. Time grows as the square of distance Interactive · How big can a cell be? Cell radius Bacterium Typical animal cell Large egg (0.5 mm) Slider is logarithmic. Spherical cell; D = 100 µm²/s is an illustrative value for a small molecule in cytoplasm (proteins diffuse more slowly). Real big cells avoid the limit by pumping fluid, streaming, or sitting still. Big cells exist, but they work around the limit. A large egg is mostly stored food and does little until it divides. Plant cells keep most of their volume as a watery vacuole, so the working cytoplasm is a thin layer. Large animals are not made of large cells but of more cells: an elephant's cells are about the size of yours. The rule of thumb is that growing up means dividing. Membranes and transport: the cell's voltage A cell is defined by its membrane, the double layer of lipids described in Biochemistry (35/43). It keeps water-soluble things out, and so every cell needs pores, channels and pumps in it for what must pass. Some molecules cross by simple diffusion, some through channels that open and close, some by carriers, and some are pumped uphill, against their concentration, using ATP. The sodium–potassium pump sends 3 sodium ions out for each 2 potassium ions in, so cells end up with high potassium inside and high sodium outside. That imbalance is a store of energy and a source of voltage. If a membrane lets one kind of ion through, that ion drifts down its concentration gradient until the electrical force pushes back with equal strength. The voltage at which this balance happens is the Nernst potential: Eion = (RT / zF) · ln([ion]out / [ion]in)at 37 °C, RT/F is about 26.7 mV; z is the ion's charge For potassium at 5 mM outside and 140 mM inside, E = −89 mV. A real nerve cell at rest sits near −70 mV, close to potassium's value because its resting membrane leaks mostly potassium; open sodium channels and the voltage moves toward sodium's +67 mV, which is how a neuron fires (see Biological Psychology, 35/20). Interactive · What voltage does each ion want? Potassium Sodium Chloride Calcium Outside concentration vs normal Typical mammalian concentrations (mM, out / in): K 5 / 140; Na 145 / 12; Cl 110 / 10; Ca 1.8 / 0.0001. Textbook values at 37 °C; they vary by cell type. Not a medical tool. Notice calcium: a 18,000-fold gradient. That is why a small leak of calcium into a cell is a loud signal, and why cells spend so much energy keeping it out of the cytoplasm. Organelles and the merger that built them Cells come in two broad plans. Bacteria and archaea (prokaryotes) are small, have no nucleus and few internal compartments, and carry their DNA in a loop in the cytoplasm. Eukaryotes, which include animals, plants, fungi and single-celled protists, are typically ten to a hundred times wider, keep their DNA in a nucleus, and have internal compartments called organelles: the endoplasmic reticulum and Golgi apparatus that process proteins, lysosomes that digest, and mitochondria, which make most of the cell's ATP. Plants and algae add chloroplasts. FeatureBacteriumAnimal cell Typical size1 to 5 µm10 to 100 µm DNAmostly one ring, in the cytoplasmseveral chromosomes, in a nucleus Energyacross the cell's own membranemostly inside mitochondria Compartmentsfew or nonemany membrane-bound Why do mitochondria look like small bacteria? Because, in the best-supported account, they were. Lynn Margulis (then Lynn Sagan) argued in a 1967 paper, rejected by many journals before it was published, that mitochondria and chloroplasts descend from free-living bacteria that were taken up by another cell and stayed. Evidence since has made this the textbook view: mitochondria have two membranes, their own small circular DNA (the human one has 37 genes), ribosomes more like bacterial ones, and they divide by pinching in two. Their closest bacterial relatives are in a group called the alphaproteobacteria; chloroplasts descend from cyanobacteria. The host was an archaeon, probably related to a group called the Asgard archaea, and in 2020 a Japanese team reported growing one such organism in the lab. How and in what order it all happened is still debated. It matters because it explains why eukaryotes could grow large and complex: each cell carries hundreds of internal power stations, so the surface-to-volume limit above no longer applies to energy production in the same way. It also explains why a cell's powerhouse has its own tiny genome. The cytoskeleton: roads, ropes and motors A cell is not a bag of soup. Three kinds of protein filaments run through it: actin (about 7 nm across) forms the cell's cortex and drives crawling; intermediate filaments (about 10 nm) are ropes that resist stretching; microtubules (about 25 nm, hollow tubes) are roads. Motor proteins walk along them carrying cargo. Kinesin takes steps about 8 nanometres long toward the end where a microtubule grows; dynein walks the other way. Each step uses one ATP. The reason cells need roads is the same diffusion law as before. Over short distances, diffusion is fine. Over long ones it is hopeless: a nerve cell in your leg can be a metre long, and its far end is made of proteins built in the cell body. Diffusion would take centuries; a motor carrying a vesicle at about a micrometre per second takes days. The crossover is where the distance equals 2D / v, which here is about 20 µm. Interactive · Diffuse or ride? Distance to deliver Across a cell, 20 µm Motor neuron, 1 cm Giraffe nerve, 1 m Diffusion in one dimension, t = L² / 2D with D = 10 µm²/s for a vesicle-sized cargo; motor at a steady 1 µm/s (the order of magnitude of fast axonal transport). Illustrative values; log scale. A metre-long nerve is a human leg or a large animal. Cell signalling: switches, not dimmers Cells also listen. A signal molecule, such as a hormone or growth factor, binds a receptor on the surface (or inside) and sets off a chain of events that ends in a change: a gene switched on, a muscle contracting, a cell deciding to divide. The largest family of receptors in humans, the G-protein-coupled receptors, number around 800 genes and are the targets of roughly a third of approved drugs. Robert Lefkowitz and Brian Kobilka won the 2012 Nobel Prize in Chemistry for them. The key idea is that cells often need yes-or-no answers, not gradual ones. One way to get a sharp response is cooperativity: when binding sites help one another, the response to a rising signal becomes steeper. The standard description is the Hill equation, where the exponent n measures steepness. With n = 1 it takes an 81-fold rise in signal to go from a 10% to a 90% response; with n = 4 only a 3-fold rise does. Haemoglobin, binding oxygen at about n ≈ 3, is the classic case, and signalling cascades can sharpen a response far more. response = [L]n / (Kn + [L]n)Hill equation; K is the signal level giving a half response; 10% to 90% takes 811/n-fold Interactive · Dimmer or switch? Steepness n Signal level, in units of K Computed from the Hill equation; the log axis runs from 0.01 K to 100 K. Cell responses have more complex mechanisms; this is the standard first model. The cell cycle, and cancer as its breakdown A dividing cell moves through a cycle: G1 (growth), S (copying DNA), G2 (checking and preparing) and M (mitosis, the division itself). Between phases are checkpoints that ask: is the DNA intact, is it fully copied, are the chromosomes attached correctly? A family of proteins, the cyclins and the kinases they activate, act as the clock. Leland Hartwell, Paul Nurse and Tim Hunt shared the 2001 Nobel Prize in Physiology or Medicine for finding these controls. Normal human cells in a dish stop after roughly 40 to 60 divisions (the Hayflick limit, Leonard Hayflick 1961), partly because their chromosome ends shorten. Cancer is what happens when a cell line escapes those controls: it ignores stop signals, divides without growth signals, avoids its own death programme and eventually can spread. It is a disease of the cell cycle, and, since each cell division copies DNA with rare errors, of accumulated mutations. A tumour of a billion cells, about a cubic centimetre, is 30 doublings from one cell (230 is just over a billion); 40 doublings give about a trillion, around a kilogram. Tumours grow far more slowly than cells cycle, because many cells die or rest. Why does cancer risk rise so steeply with age? In 1954 Peter Armitage and Richard Doll proposed that if a cell needs k separate rare changes in sequence, the incidence should rise with age raised to the power k − 1. Many adult cancers rise about like the 4th to 6th power of age, and genome sequencing later found that a typical tumour carries only a handful of "driver" mutations, often cited as two to eight. The model is simple, and later work has added others (stem-cell division counts, the immune system, the role of the tissue neighbourhood), but it explains a basic pattern. Interactive · Steps to cancer and the age curve Rate-limiting steps k Armitage–Doll: incidence ∝ agek−1. Illustrative: the curve is scaled to equal 1 at age 80. It does not diagnose, predict or compare any individual's risk. Seeing it as a breakdown of the cell cycle explains the main kinds of cancer gene. Oncogenes are accelerators stuck on; tumour suppressors are brakes that fail. The best known suppressor, p53, a protein that halts the cycle or triggers cell death after DNA damage, is mutated in about half of human cancers. Microscopy: seeing the cell Everything above was discovered by looking. A light microscope bends visible light through glass lenses, and the finest detail it can separate is set by diffraction. Ernst Abbe worked this out in 1873: the smallest resolvable distance is about the wavelength divided by twice the lens's numerical aperture (NA, the cone of light it gathers). d ≈ λ / (2 · NA)Abbe limit for light; with green light (550 nm) and the best oil lens (NA 1.4), d ≈ 196 nm About 200 nm is a hard wall for ordinary optics: mitochondria (about 0.5 to 1 µm wide) are visible, ribosomes (about 25 nm) are not. To see smaller things you need a shorter wavelength. Electrons have a wavelength of about 3.7 picometres at 100 kV, roughly 150,000 times shorter than green light, and electron microscopes have been used since the 1930s. In practice lenses limit the detail more than wavelength does, but the best reach atoms. Cryo-electron microscopy, which freezes samples in glassy ice, earned Jacques Dubochet, Joachim Frank and Richard Henderson the 2017 Nobel Prize in Chemistry. The Abbe limit was then beaten by cleverness, not by shorter waves: fluorescent labels that blink or are switched on one at a time, in techniques that earned Eric Betzig, Stefan Hell and W. E. Moerner the 2014 Chemistry Nobel. Green fluorescent protein, from a jellyfish and used to tag proteins in living cells (Shimomura, Chalfie and Tsien, Nobel 2008), turned microscopes into movie cameras for molecules. Interactive · What can this microscope resolve? Light Electrons Wavelength Numerical aperture Light: d = λ/(2 NA). Electrons: relativistic de Broglie wavelength from CODATA constants; in practice lens aberrations limit detail to about 0.1 to 0.2 nm in the best instruments, far above the wavelength. Sizes are typical values. What a graduate just sees Cells are small because surface grows as r², demand as r³, and diffusion time as r². Bigger animals have more cells, not bigger ones. A membrane with a pump and a leaky channel is a battery, and its voltage is set by concentration ratios (Nernst). Mitochondria and chloroplasts are the descendants of bacteria that moved in. Over long distances cells use motors on tracks; over short distances, diffusion is enough. Cells need switch-like responses, and cooperativity makes them. Cancer is a cell-cycle disease, built from a handful of mutations in accelerators and brakes. A microscope's resolution is limited by wavelength, and the limit can be dodged. Where it shows up next Molecular Biology (35/45): the cell's information system: DNA, RNA, protein and control of genes. Physiology (35/49): how cell voltages become nerve signals, muscle contractions and hormone responses. Plant Biology (35/53): chloroplasts, cell walls and why plants are built as they are. Microbiology (35/50): the cells that do not have a nucleus, and the world they live in. In the world: cancer screening and drugs, vaccines, stem-cell therapy, biotechnology that grows medicine in cells, and the ethics of cell lines such as HeLa, whose source, Henrietta Lacks, was treated in 1951 and whose cells were taken without her knowledge or consent. If you want to go deeper 📘 Textbook. Bruce Alberts and colleagues, Molecular Biology of the Cell (W. W. Norton, 7th edition). The standard text, thorough and beautifully illustrated. 📘 Textbook. Gerald Karp, Janet Iwasa and Wallace Marshall, Karp's Cell and Molecular Biology (Wiley, 9th edition). A more compact undergraduate alternative. 📘 Gentler. Nick Lane, Power, Sex, Suicide: Mitochondria and the Meaning of Life (Oxford University Press). The best popular account of the endosymbiosis story. Also Siddhartha Mukherjee, The Emperor of All Maladies (2010), a history of cancer. 🌐 Free course. OpenStax Biology 2e, chapters on cell structure. And MIT OpenCourseWare 7.016 Introductory Biology (Fall 2018). The takeaway A cell is a small room kept different from the outside, and its size is set by geometry and diffusion. It runs on voltage across a membrane, got its power stations by merger, moves cargo on tracks, answers signals with switches, and divides under checkpoints whose failure is cancer. Try this: look at the surface of a leaf, a thin onion skin or a spoonful of yeast in sugar-water under a magnifier or a phone lens (a cheap clip-on lens is enough). Notice the repeating units, and ask what keeps each one small. Sources & Further Reading Sources. Hooke, Micrographia (1665); Schleiden (1838) and Schwann (1839); Virchow (1855), with Remak's earlier work. Nernst potentials: standard physiology textbooks; ion concentrations as typical mammalian values. Sodium–potassium pump stoichiometry (3 Na+ out, 2 K+ in). Endosymbiosis: Margulis (as Sagan), Journal of Theoretical Biology (1967); human mitochondrial genome of 16,569 base pairs and 37 genes; Imachi et al., Nature (2020), on cultured Asgard archaea. Kinesin steps of about 8 nm: Schnitzer and Block, Nature (1997). Fast axonal transport of the order of 1 µm/s: standard neuroscience texts. GPCRs: Nobel Prize in Chemistry 2012 (Lefkowitz and Kobilka); about a third of approved drugs: Hauser et al., Nature Reviews Drug Discovery (2017). Hill equation and cooperativity: standard biochemistry texts. Cell cycle: Nobel Prize in Physiology or Medicine 2001 (Hartwell, Nurse, Hunt). Hayflick (1961). Armitage and Doll, British Journal of Cancer (1954); driver mutations: Vogelstein et al., Science (2013). p53 mutated in about half of cancers: standard reviews. Abbe (1873); electron wavelength from CODATA constants; Nobel Prizes in Chemistry for cryo-EM (2017), super-resolution (2014) and GFP (2008). HeLa: Henrietta Lacks, 1951. The diffusion numbers in the interactives are illustrative and labelled. This is not medical advice. 🌐 OpenStax Biology 2e: Cell Structure 🌐 MIT OCW 7.016 Introductory Biology 📘 Alberts et al., Molecular Biology of the Cell 📘 Lane, Power, Sex, Suicide Step 3. Physiology: How Bodies Keep Themselves Alive ---------------------------------------------------- Why: How the body keeps itself alive. A body is not a machine that runs; it is a collection of thermostats, all holding numbers steady while the world pushes. Physiology is the course that learns the set points, the sensors and the muscles of each loop. Run up a flight of stairs and four things happen without a decision. Your heart speeds up, you breathe harder, you start to feel warm, and soon you sweat. Walk back down, stand still, and all four drift back to where they began. Nothing is reading a dial; each is a loop that senses a number and pushes it back. Physiology asks how a living animal does this, organ by organ. A physiologist looks at a body and sees a set of regulated quantities (blood pressure, blood oxygen, temperature, water, sugar) and the plumbing, muscles and chemical messengers that hold them in range. This piece takes the eight ideas that stay with people who have studied it. It describes how bodies work; it is not medical advice, and nothing here replaces a clinician who knows you. The course in one box Who takes itBiology, nursing, medicine, physiotherapy, sport science and biomedical engineering students, typically in year 2 or 3. Medical schools teach it in the first year or two, often with anatomy. What it is forExplaining how organs work and what they regulate, so that what goes wrong can be understood. Comes afterCell Biology (35/44), because every organ is cells doing a job. Leads toDevelopmental Biology (35/51) and Biological Psychology (35/20) in this series, and, for medicine, pharmacology, pathology and clinical medicine. What this piece gives youYou will not be able to interpret a blood test or diagnose anyone. You will see feedback as the master idea, why a narrow vessel matters so much, why altitude is hard, how the kidney does its sum, how muscle trades force for speed, and why a mouse and an elephant live at different tempos. Easier first reads: Why You Feel Before You Think (01/01), What Exercise Actually Does (01/07) and Your Immune System, Day by Day (01/10). Homeostasis: hold the number, accept the cost In the nineteenth century Claude Bernard wrote that a complex animal lives in a stable "internal environment" even as the outside changes. Walter Cannon named the idea homeostasis in the 1920s and gave it a popular book, The Wisdom of the Body (1932). The mechanism is almost always negative feedback: a sensor measures the quantity, a control centre compares it with a set point, and an effector pushes it back toward the set point. correction = gain × (set point − measured value)bigger gain, smaller error left over; but add a delay and a big gain makes the loop swing Two things follow, and they are the heart of control theory (see Control Systems, 34/20). A loop with finite gain never removes the error completely: a body in a cold room settles slightly below its set point. And a loop with a delay between sensing and acting overshoots; push the gain too high and it oscillates, like a shower where the hot tap answers ten seconds late. Interactive · Your own thermostat Loop gainDelay (steps) A toy model, not a body: xt+1 = xt + 0.1·(Tenv − xt) + 0.1·gain·(37 − xt−delay). The environment drops from 37 to 17 at step 10. Arbitrary time units, illustrative. With gain 4 and no delay the model settles near 33, short of 37, and it settles smoothly (the exact answer is (17 + 4 × 37) ÷ 5 = 33.0). Raise the gain to 15 and the target moves to (17 + 15 × 37) ÷ 16 = 35.75, a dip of only 1.25 degrees. Add a delay of just 2 steps and gain 15 runs away completely; even gain 4 keeps ringing with a delay of 4. Real regulation keeps gain high, stacks several loops and uses fast sensors, which is why healthy bodies are steady and why tired or sick ones can overshoot. The heart and circulation: flow equals push over resistance The heart is two pumps in series. The right sends blood through the lungs; the left sends it round the body. In each beat a ventricle ejects a stroke volume; times the number of beats a minute, that is the cardiac output. At rest a healthy adult is typically around 5 litres a minute, for example 70 beats of about 70 mL. Pushing blood round the body is like pushing water through pipes: the pressure you need is flow times resistance. MAP ≈ CO × TPR  ·  resistance ∝ 1 / radius4mean arterial pressure ≈ cardiac output × total peripheral resistance · Poiseuille's law for a tube The fourth power is the striking part. Halve a tube's radius and resistance rises sixteen-fold. That is why the body steers blood by making small vessels a little narrower or wider (after a meal, more to the gut; in the cold, less to the skin), and why a small narrowing of an artery matters so much. Real vessels are elastic and blood is not a simple fluid, so Poiseuille's law is a guide, not a prediction. Interactive · Pump and pipes Heart rate (beats/min)Stroke volume (mL)Small-vessel radius vs resting (%) Resting total peripheral resistance taken as 18 mmHg per L/min (illustrative round number). Central venous pressure ignored. Resistance scales with radius−4. Not a clinical calculator. At 70 beats × 70 mL the output is 4.9 L/min and pressure comes out near 88 mmHg. Narrow the small vessels by 10% and resistance rises by about 52% (1 ÷ 0.94), pressure follows, unless the heart compensates. A fit person at rest has a bigger stroke volume and a slower rate for the same output. Lungs and gas exchange: diffusion needs area and a gradient Oxygen crosses from air to blood by diffusion alone. Fick's law says that flow of gas through a membrane is proportional to the area, the pressure difference of that gas, and inversely to thickness. So the lung is built as a huge surface a few cells thin: some 300 million tiny air sacs (alveoli) give a gas-exchange surface of tens of square metres (often quoted as about 70; estimates differ) wrapped in blood capillaries. What drives the flow is the partial pressure of oxygen. Air is 20.95% oxygen at every altitude; what changes is the total pressure. Climb and the push to get oxygen into blood shrinks. The oxygen pressure in the lung's air sacs is lower than in the air you breathe, because you subtract carbon dioxide, which is why breathing harder helps: PAO2 ≈ PIO2 − PACO2 ÷ 0.8alveolar oxygen ≈ inspired oxygen − CO₂ ÷ respiratory ratio · PIO2 = 0.2095 × (Ptotal − 6.3 kPa water vapour) Interactive · Climb a mountain Altitude (m)Alveolar CO₂ (mmHg) Standard-atmosphere pressure P = 101.325·(1 − 2.25577×10−5h)5.256 kPa; 1 kPa = 7.5 mmHg. Real pressure varies with weather, latitude and season (Everest is higher than the standard atmosphere gives). At sea level the model gives an inspired oxygen pressure of about 19.9 kPa (149 mmHg) and an air-sac value near 100 mmHg. On a 4,000 m pass the inspired oxygen pressure is about 58% of sea level. At the height of Everest, standard-atmosphere pressure leaves an inspired value about 26% of sea level, and with normal breathing the air-sac value is below zero in this model: the only way to have any oxygen at all is to breathe much harder and blow off CO2, which climbers do. Kidneys: filter everything, take back almost all The kidneys do not look for waste and pull it out. They do the opposite: about a million tiny filters in each kidney (nephrons) push a large share of blood plasma, water and small molecules, into a tube, and then the tube takes nearly all of it back. What is not taken back is urine. Normal filtration is around 125 mL a minute, about 180 litres a day, from a plasma volume of only about 3 litres: the blood is filtered some 60 times a day. urine = filtered × (1 − fraction reabsorbed)about 180 L filtered a day · about 99% reabsorbed · about 1.5 L left Because the filter is so large, a tiny change in the fraction taken back makes a large change in the output. A hormone from the pituitary, ADH (vasopressin), tells the collecting ducts how many water channels to open, and the nephron's loop, a hairpin that concentrates the fluid around it, sets how concentrated urine can get. Animals of dry habitats, such as desert rodents, have long loops and make far more concentrated urine than we can. When kidneys fail, dialysis imitates the filter; it is a hard job, which shows how much the kidney does. Interactive · A sum with a huge multiplier Normal dayDehydratedDrank a lotFiltration rate (mL/min)Fraction reabsorbed (%) Presets set illustrative reabsorbed fractions (99.2, 99.6 and 98.5%). Urine = rate × 1440 min × (1 − fraction). A healthy kidney varies this by the hour; this is not a diagnostic tool. At 125 mL/min and 99.2% reabsorbed, urine is 1.44 L a day. Move the fraction to 98.5% and it doubles to 2.7 L; to 99.6% and it halves to 0.72 L. A swing of half a percentage point in what is taken back is the whole difference between a thirsty day and a wet one. Hormones: slow messages through the blood Nerves are for speed, in milliseconds. Hormones are for broadcast and persistence: a gland releases a chemical into blood, which reaches every cell, and only cells with the right receptor respond. Ernest Starling coined the word in 1905; the hormone insulin was isolated in Toronto in 1921, and the 1923 Nobel Prize went to Banting and Macleod, who shared their money with Best and Collip. The loops are the same negative feedback as before, just slower. The hypothalamus tells the pituitary, the pituitary tells the thyroid or adrenal gland, and the hormone released tells the hypothalamus to ease off. Blood sugar is a two-hormone loop: insulin from the pancreas helps cells take up glucose after a meal, and glucagon raises it between meals. A few loops run positive, so that one event feeds itself to a conclusion: the surge of luteinizing hormone that triggers ovulation, or oxytocin in labour. How long a hormone acts is set by how fast it is cleared. Adrenaline is gone in minutes, which suits an emergency; thyroid hormone (thyroxine) has a half-life of about a week, so its level takes weeks to move and it is a slow tuner of metabolism. In general a level reaches 90% of a new steady value after about 3.3 half-lives, so a hormone with a long half-life cannot be quickly "reset" by anything. Popular "hormone balancing" claims in wellness ads usually ignore all of this: hormones are tied to feedback loops, so pushing one tends to be answered by the others. Muscle: force and speed trade off A muscle fibre is a stack of overlapping thick (myosin) and thin (actin) filaments. In 1954 two teams, Andrew Huxley with Rolf Niedergerke and Hugh Huxley with Jean Hanson, independently proposed the sliding filament model: the filaments keep their length, and myosin heads grab the actin and pull it past in repeated strokes, each powered by ATP and triggered by calcium. Muscle shortens because many tiny ratchets act in parallel and in series. Because each stroke takes time, a fast shortening muscle has fewer heads attached at once: it can pull hard or fast, not both. A. V. Hill described the curve in 1938: (F + a)(v + b) = (F0 + a) bF = load, v = shortening speed, F₀ = maximum force, a and b = constants of the muscle Power is force times speed, so it is zero at both ends (no speed when holding a maximal load, no force when moving freely) and peaks at about a third of the maximal force for typical constants. That is why a bicycle has gears: they let your legs work near the power peak. Interactive · The force–velocity trade Load (% of maximum force) Hill's equation with a = 0.25·F0 (a common textbook value; real muscles differ). Speed shown as a fraction of its no-load maximum, power as a fraction of its peak. With those constants, power peaks when the load is 31% of the maximum and the muscle is shortening at 31% of its fastest speed; the peak is about 9.5% of F0 × vmax. Raise the load to 80% and speed falls to 5% of maximum. Temperature control: a thermostat with a heater, a fan and a pipe Mammals and birds hold the core near 37 °C (set points differ by species) by balancing heat made against heat lost. The hypothalamus compares the signals from the skin and core with a set point and drives the effectors: shivering and brown fat make heat; narrowing skin vessels keeps it in; widening them and sweating let it out. Sweat cools by evaporation, which needs the air to take up more water. Fever is not a failure of the thermostat but a raised set point: you feel cold until the body reaches it. That cooling has a limit when the air is hot and wet. In 2010 Steven Sherwood and Matthew Huber argued that a sustained wet-bulb temperature of 35 °C is a theoretical limit for human survival. In 2022 laboratory work on young, healthy adults found people hitting their limit well below that, and the value depends on both temperature and humidity. The honest summary is: there is no single threshold, and older and ill people are at risk at lower values. Scaling: Kleiber's law and the pace of small animals If you double an animal's length, its mass goes up eight-fold, but the surface it loses heat through goes up only four-fold. Small animals lose heat faster for their weight, so they burn more per kilogram. In 1932 Max Kleiber compared animals from rats to cattle and found that basal metabolic rate rises not as the mass nor the surface (the 2/3 power) but about as the three-quarter power of mass. BMR ≈ 3.4 × M0.75 wattsM in kilograms · same as 70 × M0.75 kcal a day A 70 kg person comes out near 82 W (about 1,700 kcal a day), a 25 g mouse about 0.2 W, an elephant of 4 tonnes about 1.7 kW. Per kilogram, the mouse burns around twenty times what an elephant does, and its heart races at several hundred beats a minute where the elephant's beats about thirty. West, Brown and Enquist proposed in 1997 that the exponent comes from the branching of the circulatory network; others (Kozłowski, Konarzewski) disagree, and data for small birds and insects often fit 2/3 better. Whether the exponent is exactly 3/4 is still argued over. Interactive · A mouse, a person, an elephant 3/4 (Kleiber)2/3 (surface)Body mass (log scale) Both lines are anchored to 82 W at 70 kg. Heart-rate rule of thumb HR ≈ 240 × M−1/4 beats/min; real species scatter widely around it. Slider is log10 of mass in kg. Slide to a 25 g mouse: about 0.2 W under the 3/4 law, against 0.4 W on the 2/3 line, and 8.6 W per kg against 1.2 for a person. Often-repeated claims that every mammal gets the same number of heartbeats in a lifetime are only roughly true; people, with a heart rate of about 70, clearly exceed it. What a graduate just sees Everything regulated has a sensor, a set point and an effector; read any disease as a loop that has broken or been overloaded. Feedback with a delay and a lot of gain oscillates; the body needs fast sensors and several loops. Flow equals pressure over resistance, and resistance goes with radius to the fourth power, so a slightly narrower artery is a big deal. Oxygen delivery is diffusion down a pressure gradient across a very large, thin surface. The kidney filters everything and reabsorbs almost all, so tiny percentage changes make big volume changes. Hormones are slow because they are broadcast and cleared slowly; half-life sets the tempo. Muscle is a force–speed trade; maximum power sits at a modest load. Metabolic rate rises more slowly than mass, so small animals live fast. Where it shows up next Biochemistry (35/43) and Cell Biology (35/44) sit below it: ATP, receptors and ion pumps are the parts of every loop. Biological Psychology (35/20) and Control Systems (34/20) use the same feedback logic for brains and machines. Microbiology (35/50): infection is a challenge physiology answers with fever, immunity and inflammation. In the world: sport and training, medicine and nursing, high-altitude expeditions, heat-wave planning, drug dosing, and wearables that track heart rate and sleep. If you want to go deeper 📘 Textbook. Dee Unglaub Silverthorn, Human Physiology: An Integrated Approach (Pearson). A widely assigned textbook built around homeostasis and feedback. 📘 Textbook. Arthur Guyton and John Hall, Textbook of Medical Physiology (Elsevier). The classic medical-school reference, heavy on quantitative reasoning. 📘 Gentler. Bill Bryson, The Body: A Guide for Occupants (2019). A very readable tour with almost no maths. 🌐 Free course. OpenStax Anatomy and Physiology 2e, a free full textbook, and Khan Academy's Human anatomy and physiology videos. The takeaway A body stays alive by holding a few numbers steady with feedback loops, and each organ is a sensor, a pump or a filter in one of them. The equations are small (flow = pressure ÷ resistance, diffusion needs area and a gradient, force trades against speed, rate scales as mass to the three-quarters), and they explain surprisingly much of what you feel. Try this: this week, take your resting pulse for 30 seconds, then climb two flights of stairs and take it again. Then watch how long it takes to settle. That recovery time is a feedback loop made visible. (This is curiosity, not a fitness test.) Sources & Further Reading Sources. Feedback, Poiseuille's law, Fick's law, the alveolar gas equation, renal handling, the force–velocity curve (A. V. Hill, 1938) and the sliding-filament model (Huxley and Niedergerke; Huxley and Hanson, Nature, 1954) are standard textbook physiology (Silverthorn; Guyton and Hall). Kleiber (1932) and the West–Brown–Enquist (1997) debate were checked against an encyclopaedia summary. Cardiac output and GFR typical values from encyclopaedia summaries of textbook figures. The wet-bulb limit: Sherwood and Huber (2010) and Vecellio et al. (2022). Not medical advice. All interactive numbers were computed in node. 🌐 OpenStax Anatomy and Physiology 2e (free) 🌐 Khan Academy: Human anatomy and physiology 📘 Silverthorn, Human Physiology: An Integrated Approach 📘 Bryson, The Body: A Guide for Occupants Step 4. Anatomy: Where Everything Is, and Why There --------------------------------------------------- Why: The body's structure, region by region. The body is not a heap of organs but a set of rooms, pipes, levers and tubes, each shaped by the job it does. Anatomy is the course that learns the map, so that every later course has somewhere to stand. Put two fingers on the front of your wrist and you feel a hard cord moving as you wiggle your fingers. Nothing about it is magic: it is a tendon, a rope that runs from a muscle in your forearm, across the wrist, to a bone in your finger. Your fingers have no muscles big enough to grip a cup, so the muscles live up in the forearm and pull on long ropes. A surgeon, a physiotherapist and a sculptor all see the same arrangement. Most of us have never been told it. Anatomy asks the plain question first: what is where, what is it attached to, and what lies in between? The answer, once learned, makes the rest of medicine legible. A pain "under the ribs on the right" is not a vague place; it is a short list of structures. This piece explains how medicine works and is not medical advice. For a health question, see a clinician; in an emergency, call your local emergency number. The course in one box Who takes itMedicine, dentistry, nursing, physiotherapy, sport science, veterinary and biomedical students, usually in the first year. Medical schools teach it as "gross anatomy" (what the eye can see), alongside histology (tissues under a microscope), embryology (how the body forms) and neuroanatomy. Some schools use full dissection of donated bodies, some use prepared specimens, some mostly screens and models. What it is forGiving every later course a map: you cannot reason about a heart attack, an operation or an X-ray without knowing where the parts are. Comes afterCell Biology (35/44), because every structure is made of tissues, and tissues of cells. Leads toPathology (35/55), Physiology (35/49, taught beside it) and Surgery, Anaesthesia & Emergency Care (35/62). What this piece gives youYou will not be able to name the 206 bones of an adult skeleton or pass an anatomy practical. You will see the body as regions and systems, speak the language of "proximal" and "sagittal", see why a bone is a tube and an arm a bad lever, read what each kind of scan shows, and know why no body matches the atlas. Easier first reads: Physiology (35/49) and the 01 Nervous System & Body pieces; for the beam idea, Mechanics of Materials (34/11). Regions and systems: two ways to cut one body There are two ways to organise a body, and an anatomist uses both. Regional anatomy asks what is in the head, the neck, the chest, the abdomen, the pelvis and the limbs. Systemic anatomy follows one network through every region: the skeleton, the muscles, the blood vessels and heart, the airway and lungs, the gut, the urinary and reproductive organs, the nerves, the glands, the skin, and the lymph and immune organs. Regions tell you what you will meet when you open a part; systems tell you where a pipe, a nerve or a vessel is heading. A good picture of the body is the grid of both. Notice what regions have in common. The skull, rib cage and pelvis are bony containers that protect soft organs; the neck is a crowded corridor; the abdomen is the soft, packed one; the limbs are bone-and-muscle machines. Pick a region and see. Interactive · A body-region explorer HeadNeckChestAbdomenPelvisLegsArms A schematic, not a drawing to scale; textbook arrangement, and individual bodies differ (see the variation section). Not medical advice. Language and planes: say where without pointing Anatomy has a vocabulary because "above the knee" is ambiguous: above when you stand, or above when you lie down? The fix is a fixed reference pose, the anatomical position: standing, facing forward, arms at the sides, palms forward. Every direction is described from it, and left and right mean the patient's left and right, not yours. That one rule prevents operations on the wrong side. TermMeaningExample Anterior / posteriorfront / backThe sternum is anterior to the spine. Superior / inferiortoward the head / toward the feetThe heart is superior to the stomach. Medial / lateraltoward the midline / away from itThe thumb is lateral, the little finger medial. Proximal / distalnearer / farther from the trunk, along a limbThe wrist is distal to the elbow. Superficial / deepnearer / farther from the surfaceSkin is superficial to muscle. The body is also cut by three planes. A sagittal plane splits it into left and right, a coronal plane into front and back, and a transverse (or axial) plane into top and bottom. These are not just words: they are how every scan is drawn. A CT slice is a transverse plane, and by convention it is shown as if you were looking up from the patient's feet, so the left side of the picture is the patient's right. Drag the cut to see what it passes through. Interactive · Cut the body three ways TransverseSagittalCoronal Height of the transverse cut Region boundaries are rounded textbook proportions of standing height; the real levels move with body shape and breathing. Structure and function: a bone is a tube, an arm is a lever The master idea of anatomy is that shape is explained by the job. Two examples show how far this goes. First, a long bone is a tube. Bending stiffness of a rod depends on how far its material sits from the middle, in fact on the fourth power of the radius, so for a fixed amount of material a hollow tube is much stiffer than a solid rod. Bones are not simple pipes (they have a hard outer shell, a spongy interior and marrow), but the principle holds, and it is the same reason a bicycle frame and a drinking straw are tubes. If you have met a beam in Mechanics of Materials (34/11), you have met the reason. stiffness ∝ I = (π/4)(Rout4 − Rin4)second moment of area for a tube · for the same material area, stiffness ratio to a solid rod = 2x² − 1 where x = Rout ÷ rod radius Interactive · Same material, hollow Outer radius (solid rod = 1) Idealised round cross-section of equal area, ignoring bone's shell-and-sponge structure. Opening out to 1.5 times the rod radius gives 3.5 times the stiffness; to twice the radius, 7 times. Second, an arm is a lever, and a bad one. The biceps attaches to the forearm only a few centimetres from the elbow, while your hand is about 35 cm away. A lever like that gives up force to gain reach and speed: the hand sweeps a large arc while the muscle shortens a little. The price is a large muscle force just to hold something light. Fmuscle × dtendon = m g × Ltorque balance about the elbow · ignores the forearm's own weight, the angle, and the other elbow-flexing muscles Interactive · A lever with a poor ratio Weight held in the hand Tendon attachment, from the elbow A simplified one-muscle model with an illustrative 4.5 cm attachment and 35 cm to the hand, forearm horizontal. Real values differ between people and joint angles. With 5 kg in the hand and the tendon 4.5 cm from the elbow, the model needs about 380 N from the muscle, the pull of roughly 39 kg hanging from it: nearly eight times the load. That is why a heavy object held at arm's length tires you fast, and why tendons and the points where they attach are common places for overload injuries. Structure explains injury the same way it explains function: force over area, leverage, and the places where a stiff thing meets a flexible one. Seeing inside: imaging is anatomy by other means For most of history, anatomy meant cutting. In 1895 Wilhelm Röntgen found X-rays, and within months doctors were photographing bones in living patients. Since then, imaging has become a second way of doing anatomy without opening the body. Each method answers a different physical question. X-ray (radiograph): how much of a beam of X-rays gets through. Dense bone absorbs more than soft tissue, and air in the lungs absorbs least, so bone shows white and lung dark. A flat shadow of all that lies along the beam. CT (computed tomography): the same absorption, measured from many angles and combined by computer into thin transverse slices. Godfrey Hounsfield and Allan Cormack shared the 1979 Nobel Prize in Physiology or Medicine for it. Slices are reported on the Hounsfield scale, where water is 0 and air about −1000. MRI (magnetic resonance imaging): the body sits in a strong magnetic field; radio pulses make the hydrogen nuclei in water and fat signal back, and the signal differs by tissue. It shows soft tissues such as brain, muscle and cartilage far better than X-ray, and uses no ionising radiation, but it is slow and loud, and metal in or on the body can be dangerous. Paul Lauterbur and Peter Mansfield shared the 2003 Nobel Prize in Physiology or Medicine for it. Ultrasound: pulses of sound far above hearing, sent in and timed on return. The echo time gives depth, and no ionising radiation is used. Ian Donald and colleagues developed it for obstetrics in the late 1950s. The first physical law you need is attenuation: the fraction of X-rays that survive a path through material falls off exponentially with thickness. Each material has its own absorption coefficient μ. I ÷ I0 = e−μxfraction transmitted through thickness x · CT value in Hounsfield units: HU = 1000 × (μ − μwater) ÷ μwater Interactive · An X-ray shadow BoneFatAir-filled lung Thickness of the insert Total body path Linear attenuation near 60 keV, from NIST XCOM mass coefficients times density: soft tissue (water) 0.206, fat 0.188, inflated lung (density 0.26) 0.053, cortical bone 0.604 per cm. Real beams mix energies, so real scanners differ. With a 20 cm path of soft tissue, about 1.6% of the beam gets through. Replace 2 cm of it with bone and only about 0.7% does: the shadow of the bone is darker by more than half, which is exactly why a fracture is visible against muscle. On the Hounsfield scale this model reads bone at roughly +1900, fat near −90 and inflated lung near −740, in line with the ranges clinicians learn (the actual values depend on the scanner's energy). Ultrasound runs on another law: distance is speed times time, halved for the round trip. Scanners assume a speed of sound of about 1540 m/s in soft tissue. depth = c × t ÷ 2c ≈ 1540 m/s in soft tissue · wavelength = c ÷ frequency · loss ≈ 0.5 dB per cm per MHz each way Interactive · Echo time to depth Echo return time Probe frequency Assumes 1540 m/s and the 0.5 dB/cm/MHz rule of thumb (applied to the round trip). Real tissues vary; this is physics practice, not a clinical tool. An echo at 130 µs came from 10.0 cm deep. The trade-off is built into physics: a high frequency has a short wavelength and sees fine detail (a 10 MHz probe has a wavelength near 0.15 mm), but it is absorbed much faster, so shallow structures such as tendons and the thyroid use high frequencies and deep ones such as the liver and kidneys use lower ones, typically a few megahertz. Variation: no body is the textbook body An atlas shows one idealised body, and no body matches it. Arteries branch in more than one common pattern; some people have an extra rib or an extra muscle, and some lack a muscle most people have. The palmaris longus is the classic case: the slim tendon you can sometimes see at the centre of the wrist when you touch thumb to little finger and flex. Reviews report that it is missing in roughly 5% to 25% of people, depending on the population studied, and its absence does not weaken grip. A variant is not a disease. It matters because a surgeon who expects a structure that is not there, or a vessel in a different place, can make a serious mistake. It also matters who the "textbook body" is. Historical atlases drew a typical adult male of one ancestry; modern teaching works to show women, children, older people, and a range of skin tones and body sizes, and to teach that "normal" is a range. Interactive · How often does someone lack it? Share of people lacking the muscle People in the group The range is the span reported across populations, not a single measurement; the binomial arithmetic treats people as independent. Illustration only. In a class of 30 where 15% lack it, about 4.5 people do, and the chance that nobody does is under 1%. Variation is not the exception in anatomy; it is the rule, and the point of dissection and scanning many bodies is to learn the range. From Vesalius to body donation: whose bodies teach For fourteen centuries the authority on human anatomy was Galen, a physician of the Roman Empire whose dissections were of animals, notably Barbary macaques, because human dissection was not done in his world. In 1543 a 28-year-old Flemish anatomist, Andreas Vesalius, published De humani corporis fabrica in Basel after dissecting human bodies himself at Padua. He showed that the lower jaw is a single bone and that humans lack a network of vessels (the rete mirabile) at the base of the brain that sheep have, and he admitted he could not find the pores Galen said connected the heart's ventricles. The 13th-century physician Ibn al-Nafis had already argued that blood passes from the right side of the heart through the lungs rather than through the wall between the ventricles, and William Harvey published the full circulation in 1628. Anatomy grew by correcting itself, one body at a time. The history has a cost. Bodies for study came from executed criminals (in Britain, the Murder Act 1752 allowed dissection of executed murderers), then from graves. Demand outstripped supply, and the West Port murders in Edinburgh in 1828 helped push Parliament to the Anatomy Act 1832, which let anatomists use donated bodies, in practice often the unclaimed poor. In the United States the bodies of the enslaved and burial grounds of Black communities were targeted; in one studied case, 80% of the bodies dissected at Transylvania University in the 1830s and 1840s were African American. In the 20th century, some of the plates of the celebrated Vienna Pernkopf atlas were later found to have used executed political prisoners of the Nazi regime, which led to long debate about whether such images should be used at all. Today, medical schools in many countries rely mainly on bodies donated with informed consent, and learners are asked to treat the donor as a first teacher. Rules still differ by country, and some places still use unclaimed bodies. Anatomy teaches the body and, in the same breath, the ethics of how we came to know it. What a graduate just sees Every body has regions and systems, and a problem in one place can be read through the pipes that pass through it. "Left" and "above" are always fixed to the standing patient; the three planes (sagittal, coronal, transverse) are the grammar of every scan. A bone is a tube because stiffness grows with the fourth power of the radius. An arm is a lever that trades force for reach, so even small loads mean large muscle forces. Bone is white on an X-ray because of exponential absorption, not paint. Ultrasound turns time into depth; high frequency gives fine detail but only near the surface. Variation is normal; the atlas body is a draft, not a rule. Our knowledge of anatomy has an ethical history, and consent matters. Where it shows up next Physiology (35/49) is the same body in motion: the pump, the pipes and the filter you have just placed on the map. Pathology (35/55) describes how structures are damaged, which only makes sense once you know what normal looks like. Surgery, Anaesthesia & Emergency Care (35/62): an operation is anatomy applied, layer by layer and vessel by vessel. In the world: radiographers and sonographers, physiotherapists, athletic trainers, tattooists who know where the nerves are, forensic scientists, and illustrators. If you want to go deeper 📘 Textbook. Richard Drake, A. Wayne Vogl and Adam Mitchell, Gray's Anatomy for Students (Elsevier). The standard clinically oriented first-year text. 📘 Textbook. Keith Moore, Arthur Dalley and Anne Agur, Clinically Oriented Anatomy (Wolters Kluwer). A detailed reference with clinical notes throughout. 📘 Gentler. Bill Bryson, The Body: A Guide for Occupants (2019). A friendly tour with almost no jargon. 🌐 Free course. OpenStax Anatomy and Physiology 2e (free textbook) and Khan Academy: Human anatomy and physiology. For real slices, the US National Library of Medicine's Visible Human Project. The takeaway Anatomy is the map: regions and systems, a fixed language of direction and planes, and the habit of asking why a structure has the shape it has. Shape follows job (a tube for stiffness, a lever for reach), scans are physics applied to that map, every body is a variation on it, and the knowledge came from real people whose bodies deserve respect. Try this: this week, look at your own forearm and wrist. Make a fist, feel the tendons on the inside of the wrist, wiggle your fingers and feel the muscles move in the forearm, not in the fingers. Check for the palmaris longus tendon (it may simply not be there). You are looking at ropes and levers, not magic. Sources & Further Reading Sources. Anatomical terms, planes and the textbook layout are standard (Gray's Anatomy for Students; Moore, Clinically Oriented Anatomy). Vesalius, Galen's macaque dissections, the Fabrica (1543), the Murder Act 1752, the Anatomy Act 1832 and the West Port murders, Hounsfield and Cormack (Nobel 1979), Lauterbur and Mansfield (Nobel 2003), Ian Donald's obstetric ultrasound, the CT display convention, the 1540 m/s assumption, the palmaris longus ranges, the Pernkopf atlas and the Transylvania University figure were checked against Wikipedia articles and their sources. Röntgen's discovery (1895). Attenuation coefficients computed from NIST XCOM mass coefficients at 60 keV (physics.nist.gov/PhysRefData/XrayMassCoef) times assumed densities. The lever and tube numbers are textbook mechanics, computed in node. Not medical advice. 🌐 OpenStax Anatomy and Physiology 2e (free) 🌐 Visible Human Project (US NLM) 📘 Drake, Vogl and Mitchell, Gray's Anatomy for Students 📘 Bryson, The Body: A Guide for Occupants Step 5. Genetics: Inheritance, Chance and What Genes Can't Tell You ------------------------------------------------------------------- Why: Inheritance, and the genetics behind disease. From Mendel's peas to millions of genomes: how traits pass on, why carriers hide, what a polygenic score really predicts, and the history we should not repeat. Your eyes are not exactly your mother's or your father's. Your sister, with the same two parents, looks different from you again. Some families carry a rare illness across generations that skips whole branches. And yet a pea plant, a fruit fly and a person all obey the same few rules of inheritance, written down first by a monk with a garden. Genetics is the science of that regularity and of what breaks it. It began as arithmetic on counted plants, grew into a map of chromosomes, and is now the statistics of millions of people's genomes. It is also the subject whose misuse has done real harm, so this piece tells that history too. The course in one box Who takes itEvery biology, biochemistry and pre-medical student, usually in year 2. Medical courses reuse it for inherited disease and risk. Programmes in the Vision and Change tradition (AAAS and NSF, 2011) and courses such as MIT's 7.03 teach it as the link between molecules and populations. What it is forPredicting how traits pass between generations, and telling what a gene can and cannot explain. Comes afterMolecular Biology (35/45) and Statistics (34/05). Leads toEvolution (35/47), Developmental Biology (35/51) and Bioinformatics & Genomics (35/52). What this piece gives youYou will not be able to run a genetic analysis. You will be able to work a Punnett square, explain why carriers vastly outnumber the affected, say what a polygenic score does and does not tell you, and recognise bad arguments about human groups. Easier first reads: Nature, Nurture, and You (02/04), Personality & Individual Differences (35/26) for what heritability means, and Probability: How to Not Fool Yourself (12/03). It describes how things work and is not medical advice. Mendel, and why he was lucky Gregor Mendel, an Augustinian friar in Brno (then Brünn, now in the Czech Republic), crossed pea plants for years and counted the offspring. He presented the results in 1865 and published them in 1866. His key choice was to look at single, clear-cut traits: round or wrinkled seed, yellow or green. Cross a pure round line with a pure wrinkled one and every child is round. Cross those children with each other and about a quarter of the grandchildren are wrinkled again. His counts for seed shape were 5,474 round and 1,850 wrinkled, a ratio of 2.96 to 1. From such counts came two ideas. Segregation: each plant carries two copies of each factor (we say alleles of a gene), and each pollen or egg cell gets just one. Independent assortment: factors for different traits are shuffled separately. A dominant allele shows when present; a recessive one shows only with two copies. Dominant does not mean common or healthy. It is a statement about how the two copies combine. He was lucky in ways that are worth knowing. Each trait he chose depends mostly on one gene, with one allele fully dominant, which is rare for traits in general. He started from pure-breeding lines. And the trait pairs he crossed were not tightly linked on a chromosome, which would have blurred his ratios. Nobody read the work: it was rediscovered in 1900 by Hugo de Vries, Carl Correns and Erich von Tschermak. In 1936 the statistician R. A. Fisher noted that Mendel's counts sit closer to the ideal ratios than chance usually allows. Historians still argue about why (stopping a count at a convenient point, rounding, an overeager helper). Nobody has shown the laws are wrong; thousands of later crosses confirm them. Sampling noise is the lesson: try it. Interactive · Draw offspring Number of offspring Draw againUse Mendel's 7,324 A random draw from the expected probabilities (a seeded generator, so results repeat until you press Draw again). χ² compares observed with expected; the p-value is how often chance alone would look at least this far from the ideal. At 8 offspring a "3 to 1" cross often looks like 5 to 3 or 7 to 1. At several thousand it settles close to 3 to 1. A p-value near 1 means suspiciously perfect, a very low one means something other than chance is going on, and most real crosses sit in the middle. That is also the first lesson of Statistics (34/05): a ratio from a small sample is a rumour. Chromosomes and linkage Genes live on chromosomes. In 1902 and 1903 Walter Sutton and Theodor Boveri noticed that chromosomes pair and split in the same pattern as Mendel's factors. In 1910 Thomas Hunt Morgan, at Columbia, found a white-eyed male fruit fly among red-eyed ones, and traced the trait to the X chromosome. Humans have 23 pairs: 22 pairs of autosomes and the sex chromosomes (XX or XY). Shuffling whole chromosomes alone can give one person about 223, over 8 million, different gametes. Genes on the same chromosome travel together, except when crossing over swaps stretches between the two copies. The further apart two genes are, the likelier a swap between them. In 1913 Alfred Sturtevant, then a student in Morgan's lab, turned that into the first genetic map: 1% recombinant offspring means 1 map unit, now called a centimorgan. Genes 50 or more centimorgans apart look unlinked, as if on different chromosomes. That is the situation Mendel stumbled into. r = ½ (1 − e−2d)Haldane's map function: recombinant fraction r from map distance d in morgans · small d gives r ≈ d · large d gives r → ½ Interactive · Linked or not? Distance between two genes For a parent with alleles AB on one chromosome and ab on the other. Parental types AB and ab are equally common; recombinants Ab and aB share the rest. Mutation: where new variants come from Every allele began as a copy that went slightly wrong (see Molecular Biology, 35/45). A mutation can swap one letter, delete or insert some, or repeat a stretch many times. It matters in the germline (eggs and sperm, so it can be inherited) or in a somatic cell (which affects only that body, as in a cancer). In humans the rate is about 1.1 to 1.2 × 10−8 per letter per generation, so a child has roughly 60 to 70 new mutations that neither parent carried. Most come from the father, and the count rises with his age, by about two for every extra year. Since only some 1 to 2% of the genome codes for protein, a typical child has on the order of one new mutation in the coding part, and most of those are harmless. Examples show the range. Sickle cell disease is one letter changed (Glu6Val). The commonest cause of cystic fibrosis is a three-letter deletion that removes one amino acid, F508del. Huntington's disease is a stretch of CAG repeats that has grown too long. The ability to digest milk as an adult, lactase persistence, comes from changes next to the lactase gene that stop it switching off, and it arose independently several times, in Europe and in East Africa. And mutation is random with respect to need. In 1943 Salvador Luria and Max Delbrück showed that resistance in bacteria appears before the bacteria meet the virus that selects it, not because of it (Nobel Prize, 1969). That is the foundation of Evolution (35/47). Interactive · New mutations in one child Mutation rate per letter per generation New mutations = rate × 2 × 3.1 billion letters. Coding share taken as 1.5% of the genome (an approximate round figure). Measured counts are a little lower (60 to 70) because some regions cannot be read reliably. Hardy–Weinberg: why carriers vastly outnumber the affected In 1908 the English mathematician G. H. Hardy and the German physician Wilhelm Weinberg independently pointed out something simple. If mating is random and nothing pushes allele frequencies around, they stay put, and the three genotypes appear in fixed proportions. For two alleles with frequencies p and q (p + q = 1): p² + 2pq + q² = 1two copies of the common allele · one of each (carriers) · two copies of the rare allele (affected, if recessive) Take a recessive condition seen in 1 birth in 2,500. Then q² = 1/2,500, so q = 0.02. Carriers are 2pq = 2 × 0.98 × 0.02, about 3.9%, roughly 1 person in 25. A rare recessive allele hides almost entirely in healthy carriers, which is why such conditions seem to "come out of nowhere" in families with no history. (Cystic fibrosis in some European-ancestry populations is close to this; figures vary by population.) Sickle cell shows the opposite lesson: where malaria is common, carrying one copy protects against severe malaria (A. C. Allison, 1954), so the allele stays frequent in parts of Africa, the Mediterranean, the Middle East and India. When relatives have children, rare alleles meet more often. With an inbreeding coefficient F, affected children are q² + Fpq. First cousins' children have F = 1/16, which here raises the risk from 1 in 2,500 to about 1 in 615. Interactive · Hardy–Weinberg Frequency of the rare allele, q Inbreeding coefficient, F Genotypes: AA = p² + Fpq, Aa = 2pq(1 − F), aa = q² + Fpq. Assumes a simple recessive trait; illustrative. The same arithmetic undid an old idea. In 1917 R. C. Punnett pointed out that preventing carriers of a rare recessive from breeding changes the allele frequency very slowly: removing every affected person's genes each generation takes the frequency from q to q/(1 + q), so the time to halve a frequency of 1% is about a hundred generations. Slow, and costly in human lives. We will meet the consequences in a moment. Complex traits and polygenic scores Most traits people care about, such as height, blood pressure and risk of common illnesses, are not Mendelian. In 1918 Fisher showed that many genes each adding a small effect make a smooth bell curve, as dice do when you add up many. This is the polygenic picture, and sequencing has confirmed it. The largest study of height (Yengo and colleagues, 2022, in Nature, with about 5.4 million people) found 12,111 common variants linked to height. Together they predict about 40% of the variation in people of European ancestry. Twin studies put the heritability of height around 80%, so there is still a gap, and it is partly rare variants and effects not yet measured. A polygenic score adds up a person's copies of the variants, each weighted by its estimated effect. It is a statistic about populations, with four limits. First, it predicts a range, not a destiny: even with a good score, many individuals fall far from the prediction. Second, it works much less well in people whose ancestry differs from the studied group: studies of this transfer problem (Martin and colleagues, 2019) found scores several times less accurate in people of African ancestry than in Europeans, because most data come from Europe. Third, scores for behavioural traits such as years of schooling pick up family environment as well as direct genetic effects, and shrink when siblings are compared. Fourth, genes work through environments: the same score predicts different outcomes in different places and times. Interactive · Many small effects, one imperfect score Number of genes shaping the trait Heritability (share of variation from genes) Share of the genes your score includes 2,000 simulated people with a fixed random seed. Each gene has two equally common alleles of equal size. Trait = genes + environment, scaled so that the genes' share of variance is your heritability. An invented model for intuition, not data. Notice what the numbers say. Set heritability to 100% and one gene gives three lumps; 40 genes give a bell. At 60% heritability and a score holding half the genes, the score explains about 30% of the variation, a correlation near 0.55, and the top tenth by score are only moderately enriched for the top tenth by trait. That is the best case in a model with no ancestry gap. Human genetic variation, told correctly Any two people share about 99.9% of their DNA; a typical genome differs from the reference at roughly four to five million sites (1000 Genomes Project, 2015). Where does the difference sit? In 1972 Richard Lewontin analysed protein markers across populations and found that about 85% of the variation lay within local populations, and only a small remainder between groups. Later work with hundreds to hundreds of thousands of markers, from Noah Rosenberg and colleagues in 2002 (377 markers) onwards, found the same broad pattern: most variation, in the region of 85 to 95% depending on the markers and populations, is within populations. Two further facts keep this honest. First, A. W. F. Edwards pointed out in 2003 that combining many markers lets you place a person's ancestry accurately, which is why ancestry tests work. Both statements are true: groups can be told apart statistically, and individuals within a group differ far more than the group averages do. Second, variation is mostly continuous and geographic, not boxed. Africa holds more genetic diversity than the rest of the world, because modern humans spent longest there; people elsewhere descend from smaller groups that left, and diversity falls with distance from Africa. Traits that look like clear categories often are not. Light skin evolved more than once, with different genes in Europe and East Asia. Sickle trait follows malaria, not "race". Lactase persistence arose several times. Genes also depend on environment: phenylketonuria is a recessive inherited condition whose harm can be prevented by diet, found by newborn screening. Heritability, as Personality & Individual Differences (35/26) explains, is a population statistic in one environment; it does not measure how fixed anything is, and it cannot be compared between groups living in different conditions. Interactive · Predict, then reveal Your guess: share of all human genetic variation that lies within a single local population Reveal Measured: about 85% (Lewontin 1972); later genome-wide studies roughly 85 to 95%, depending on markers and populations. The history of eugenics Francis Galton, Darwin's cousin, coined eugenics in 1883 for the idea that society should improve its heredity by encouraging some people to have children and discouraging others. It spread around the world in the early twentieth century, across the United States, Britain, Germany, the Nordic countries, Latin America, Japan and beyond, and respectable scientists and reformers backed it. It rested on three mistakes: it treated complex traits as simple Mendelian ones, it read social disadvantage as inherited defect, and it assumed some people's lives were worth less. It was also racist in practice. Where and whenWhat happened United States, from 1907Indiana passed the first sterilisation law in 1907; many states followed. Buck v. Bell (1927) upheld Virginia's law, with Justice Holmes writing that "three generations of imbeciles are enough." Estimates of people sterilised under such laws are in the tens of thousands. Nazi Germany, 1933 to 1945A 1933 law led to the sterilisation of over 400,000 people. The "euthanasia" programmes killed many thousands of disabled people, and the same personnel and methods fed into the Holocaust. Sweden, 1934 to 1976A government inquiry in 2000 estimated about 21,000 people forcibly and about 6,000 coercively sterilised, mostly women. Alberta, Canada, 1928 to 1972About 2,800 sterilisations under the province's Sexual Sterilization Act. Japan, 1948 to 1996The Eugenic Protection Law led to about 25,000 forced sterilisations; Japan's Supreme Court held it unconstitutional in July 2024. Peru, 1995 to 2000A national family-planning programme sterilised about 25,000 men and some 330,000 women according to a 2002 report; coercion of poor and Indigenous women was alleged and investigated. India, 1975 to 1977During the Emergency, millions of vasectomies were carried out, many under pressure. Many early critics were geneticists, as the Punnett argument above shows. The practice fell out of favour after 1945, though states went on sterilising for decades. Today's questions are subtler: prenatal screening, which is a choice, but one made in a society that treats some lives as lesser; and companies offering embryo selection using polygenic scores, which the European Society of Human Genetics has called unproven. Disability-rights writers and geneticists alike warn that the old logic can return in softer forms. The lesson of the history is not that genetics is dangerous. It is that a statistic about populations is not a verdict on a person. What a graduate just sees Ratios from small samples are noisy. A "3 to 1" cross of 8 offspring is not a failed law. Dominant describes how alleles combine, not how common or healthy they are. Genes close together travel together; genes far apart behave as if on different chromosomes. Most new mutations are neutral, and they arise without regard to need. Rare recessive alleles hide in carriers, who are far more common than the affected. Most traits are polygenic: many small effects, a bell curve, and a score that predicts a range for a population, not an outcome for a person. Most human genetic variation is within populations, and it is geographic and continuous, not boxed. Heritability is about variation in a population in an environment; it is not fate and not a ranking of groups. Eugenics was mainstream science plus prejudice, and it is a reason to be careful with every new use of genetic prediction. Where it shows up next Evolution (35/47): Hardy–Weinberg is the null model; selection, drift, migration and mutation are the forces that break it. Developmental Biology (35/51): the same genes act at different times and places; mutations that change regulation change bodies. Bioinformatics & Genomics (35/52): genome-wide studies, polygenic scores and ancestry inference are statistics on exactly these ideas. Medicine and the clinic: carrier screening, newborn screening, inherited cancer risk, pharmacogenetics and genetic counselling. Agriculture: crop and livestock breeding, from Mendel's peas to marker-assisted selection. If you want to go deeper 📘 Textbook. Anthony Griffiths and others, Introduction to Genetic Analysis (Macmillan/W. H. Freeman, 12th edition, 2020). A standard course text, strong on problem-solving. 📘 Textbook. Benjamin Pierce, Genetics: A Conceptual Approach (Macmillan, 7th edition, 2020). Clearer on the concepts, with many worked problems. 📘 Gentler. Adam Rutherford, A Brief History of Everyone Who Ever Lived (2016) and Control: The Dark History and Troubling Present of Eugenics (2022); Kathryn Paige Harden, The Genetic Lottery (Princeton, 2021); Siddhartha Mukherjee, The Gene (2016). 🌐 Free course. MIT OpenCourseWare 7.03 Genetics (Fall 2004), and the genetics chapters of OpenStax Biology 2e. The takeaway Inheritance is arithmetic on pairs of alleles: segregation, assortment, linkage and sampling noise. New alleles arise by random mutation; rare recessives hide in carriers; most traits come from many genes of small effect plus environment, so predictions are statistics about groups. People differ genetically, but mostly within populations and along gradients, and the story of eugenics shows what happens when a statistic is mistaken for a verdict. Try this: next time you read "X% heritable" or "the gene for Y", ask what population, what environment, and whether the finding is about one person or about a group. Sources & Further Reading Sources. Mendel, Versuche über Pflanzen-Hybriden (1866), seed-shape counts 5,474 and 1,850; Fisher, Annals of Science (1936); Hardy (1908) and Weinberg (1908); Sturtevant (1913); Haldane (1919) map function; Luria and Delbrück (1943); Allison (1954); Punnett (1917); Yengo and others, Nature 610 (2022); Martin and others, Nature Genetics (2019); Lewontin (1972); Edwards (2003, on "Lewontin's fallacy"); Rosenberg and others, Science (2002); 1000 Genomes Project, Nature (2015); Kong and others, Nature (2012) on mutation rates and father's age; Galton (1883); Buck v. Bell, 274 U.S. 200 (1927); Swedish government inquiry (2000) and other counts as summarised in encyclopaedic overviews, with estimates varying; Japan Supreme Court ruling, July 2024. Describes how things work and is not medical advice. 🌐 MIT OCW 7.03 Genetics 📘 Griffiths and others, Introduction to Genetic Analysis 📘 Harden, The Genetic Lottery 📘 Rutherford, Control Step 6. Pathology: How Bodies Go Wrong, in Order ------------------------------------------------ Why: Inflammation, repair, tumours and the mechanisms of disease. Disease has a small number of ways of happening: cells are hurt, the body reacts, it repairs or scars, blood flow fails, or growth escapes control. Pathology is the course that learns those few patterns, so the thousands of named diseases stop looking unrelated. You bang your shin on a table. Within minutes it is red, warm and swelling; within days it is yellow-green; within weeks it is gone. Nothing was prescribed. A tiny, ordered programme of damage, response and repair ran by itself, and doctors can name every step. That programme is the root of most pathology. The same few moves (injury, inflammation, healing, a blocked vessel, a growth that will not stop) explain a surprising share of what medicine treats, from a heart attack to a scar to a tumour. A pathologist looks at a piece of tissue and asks which move happened, and what it did. This piece explains how medicine works and is not medical advice. For a health question, see a clinician; in an emergency, call your local emergency number. The course in one box Who takes itMedicine, dentistry, veterinary, nursing and biomedical students, usually in the second year. "General pathology" covers the shared mechanisms (this piece); "systemic pathology" then goes organ by organ. Programmes in the UK and much of the world teach it inside an integrated course; US medical schools usually have a separate pathology block. What it is forExplaining the mechanisms of disease, so that signs and symptoms can be traced to what happened in the tissue. Comes afterCell Biology (35/44), Physiology (35/49) and Anatomy (35/54): you need normal before abnormal. Leads toClinical Reasoning & Diagnosis (35/59), and with Pharmacology (35/56) the whole clinical years; Immunology (35/57) goes deeper into the response described here. What this piece gives youYou will not be able to read a biopsy slide. You will see why a clot kills tissue, why some injuries heal clean and others scar, why a tumour is silent for years then seems to appear overnight, and why a single sample can miss a cancer. Easier first reads: Cell Biology (35/44), Physiology (35/49) and Your Immune System, Day by Day (01/10). Cell injury and death: reversible, then not Cells live on a tight budget of oxygen, energy and intact membranes. Hurt them (too little blood, a toxin, an infection, heat, a genetic fault) and they pass through two stages. At first the injury is reversible: the cell swells, its energy pumps slow, fat may build up inside; take the cause away and it recovers. Past a threshold, the injury is irreversible: membranes break, the cell's own enzymes leak, and the cell dies. The idea that matters is the threshold, because the whole of emergency medicine is a race to remove the cause before it is crossed. Cells die in two main ways. Necrosis is death by accident: the cell bursts or is digested, spills its contents, and the body reacts with inflammation. This is what happens in a heart attack. Apoptosis is death by design, a tidy self-dismantling that leaves no mess. It sculpts the fingers of an embryo and removes damaged cells every day. John Kerr, Andrew Wyllie and Alastair Currie named it in 1972; the 2002 Nobel Prize in Physiology or Medicine went to Sydney Brenner, Robert Horvitz and John Sulston for finding the genes that control it. Too little apoptosis lets damaged cells survive (cancer); too much kills useful ones. The threshold is different for each tissue because each has a different appetite for oxygen. Brain cells starve in minutes; heart muscle in tens of minutes; resting skeletal muscle can tolerate hours. Restoring blood is not purely good, either: when blood returns to starved tissue it can bring a second wave of damage called reperfusion injury, one reason treatment timing is delicate. Interactive · The point of no return BrainHeart muscleSkeletal muscle Minutes without blood A cartoon with round, illustrative tolerance times (5, 30 and 180 minutes at which half the cells have died). Real thresholds depend on temperature, blood left in the tissue and the cell type; this is not a clinical rule. On the cartoon, heart muscle starved for 20 minutes has lost about 12% of its cells; at 30 minutes, half. That steep S-shape is the point: damage is slow, then fast, and the middle is where saving tissue is possible. Inflammation: the fast response, and what happens when it does not stop Inflammation is the body's first reply to injury or infection. The Roman writer Celsus listed four signs: redness (rubor), swelling (tumor), heat (calor) and pain (dolor); loss of function is usually added as a fifth. Each has a cause you can follow. Local blood vessels widen (redness, heat) and become leaky, letting fluid and defence cells out (swelling); chemical signals stimulate nerves (pain). It looks like damage but it is defence. Acute inflammation lasts hours to days. Neutrophils, short-lived cells that arrive first, flood in within hours to kill microbes and eat debris. Macrophages follow within a day or two, clear what is left, and signal that repair should begin. If the cause is removed, the inflammation resolves. If it is not (a persistent infection, an irritant such as silica or tobacco smoke, or the immune system attacking the body's own tissue) it becomes chronic: macrophages and lymphocytes accumulate, tissue is damaged and rebuilt at the same time, and scar slowly replaces working tissue. Chronic inflammation is part of the story of conditions as different as rheumatoid arthritis, hepatitis and atherosclerosis. The way to remember it: acute inflammation is a fire brigade, chronic inflammation is a fire that was never put out. Interactive · Who is in the room, hour by hour ResolvesNever stops Time since injury A schematic of the order of arrival, not measured counts. Peak times (neutrophils about 12 hours, macrophages about 3 days) are round numbers; they vary with tissue and cause. Healing and scarring: regenerate or patch After injury a tissue has two choices. Regeneration replaces like with like. Repair plugs the gap with a quicker, cruder material, collagen, which is scar. Which one you get depends on whether the tissue's cells can divide and whether its framework survived. Skin surface, gut lining and blood cells are replaced constantly. The liver is quiet but can regrow. Heart muscle and most brain neurons almost never do. Wound healing runs in overlapping phases: bleeding stopped by a clot, inflammation, growth of new tissue and vessels, then months of remodelling. Scar tissue is never quite as good as what it replaces. In skin, collagen strength reaches only around 20% of normal after three weeks and climbs to at most about 80% after a year. A scar does not sweat, grow hair or stretch like skin, and in organs scar can be a disease in itself: cirrhosis of the liver, a scar in a heart after infarction, fibrosis of the lung. In some people scar overgrows its wound (keloid). Interactive · Guess first: regenerate or scar? Skin surfaceLiverBroken boneHeart muscleBrain neuronsCut nerve I think it regeneratesI think it scars General outcomes; real healing depends on the size of the injury, age, blood supply and health. Haemodynamics: clots, emboli and infarcts Haemodynamics asks what happens when blood flow goes wrong. Blood is built to clot at a cut, and the same machinery can fire where it should not. A thrombus is a clot that forms inside an intact vessel. An embolus is a lump (usually a clot) that breaks away and travels until it lodges. An infarct is the area of dead tissue behind a blockage. A clot in a leg vein can travel through the heart to the lungs; a clot on a plaque in a heart artery can starve a patch of heart muscle. Three conditions favour a clot: slow or turbulent flow, a damaged vessel lining, and blood that clots too easily. They are called Virchow's triad, although the German pathologist Rudolf Virchow never proposed it as a triad: the name was attached to him decades after he died. How much tissue dies depends on where the block is and what backup exists. A block high in the tree starves everything downstream. A block low in the tree affects a small territory. Tissues with a rich web of side connections (the hand, the gut arcades) survive a blockage that would kill tissue supplied by a single "end" artery. Interactive · Block a branch TrunkLeft branchRight branchTwig 1Twig 2Twig 3Twig 4 No backup (end artery)Some backupRich backup A toy tree with four equal territories; "backup" rescues an illustrative 0, 50 or 90% of downstream tissue. Real organs are uneven. Block the left branch of an end-artery organ and half the tissue is at risk; with some backup, a quarter dies; with rich backup, 5%. Same clot, very different outcomes: pathology is about location as much as cause. Neoplasia: how cancers start and grow A neoplasm (new growth) is a clone: the descendants of a single cell that picked up changes in its DNA allowing it to ignore the normal limits. Cancer genes come in two kinds, oncogenes that jam the accelerator and tumour suppressors that are the brakes. In 1971 Alfred Knudson, studying a childhood eye cancer (retinoblastoma) that occurs in both inherited and non-inherited forms, proposed that a suppressor gene needs both copies to be lost, a "two-hit" idea that explained why inherited cases appear earlier: one hit is already there at birth. In 2000 Douglas Hanahan and Robert Weinberg organised the traits a cancer needs (self-sufficient growth signals, ignoring stop signals, evading cell death, endless replication, a blood supply, and invasion) as the "hallmarks of cancer". A benign tumour grows in place and stays encapsulated; a malignant tumour invades neighbours and can spread. Benign does not always mean harmless: a benign mass in a skull or against a vessel can do harm by pressure alone. Cancers are common because many steps are needed, and these take time: in 1954 Peter Armitage and Richard Doll noticed that cancer rates climb as a steep power of age, as expected if several separate changes must accumulate. If, say, six steps were needed, the rate at 80 would be 32 times the rate at 40. The IARC estimated about 20 million new cancer cases and 9.7 million cancer deaths worldwide in 2022, and that roughly one person in five develops cancer in a lifetime. Tumours also grow on a clock that fools intuition. A tumour that starts as one cell must double about 30 times to reach around a billion cells, roughly a centimetre across and about a gram: the size at which it may first be found. Another 10 doublings would make it about a kilogram. Most of a tumour's life is silent, and the last stretch is quick. Interactive · The long silent start Doubling time Steady exponential growth, the textbook simplification. Real tumours slow as they outgrow their blood supply, shed cells, and vary from days to years in doubling time. Not a prediction for anyone. At a doubling time of 100 days, the model takes about 8.2 years to reach a centimetre and just 2.7 years more to reach a kilogram. That is why early detection matters so much, and why growth that "appeared overnight" was usually years in the making. Spread: invade, travel, settle The thing that makes cancer lethal is rarely the first lump; it is metastasis, the spread to distant organs, which accounts for most cancer deaths. A cancer cell must break from its neighbours, invade local tissue, enter a lymph channel or blood vessel, survive the journey and the immune system, stop in a distant small vessel, leave it, and then grow in a foreign tissue. Every step is hard, so the great majority of cells that escape die. The few that succeed do not choose randomly: the 19th-century "seed and soil" idea of Stephen Paget, that certain cancers prefer certain organs, still frames the discussion. Lymph nodes, liver, lung, bone and brain are common destinations. This is why staging exists. The TNM system records the size of the Tumour, spread to lymph Nodes and Metastasis, and tells clinicians (and patients) how far the disease has gone, which tends to predict outcome better than the tumour's appearance alone. The autopsy and the biopsy: pathology as evidence Pathology is also a method. In 1761 Giovanni Morgagni, in De sedibus et causis morborum, matched the symptoms his patients had in life with the damage he found after death, and so made disease a matter of places in the body. In 1858 Rudolf Virchow's Cellular Pathology pushed the place of disease down to the cell. Today the same tools run in two forms. An autopsy examines a whole body after death to find out what happened. A biopsy takes a piece of living tissue (with a needle, a punch or surgery), which is cut thin, stained and read under a microscope; more tests then look at proteins and genes. Both are still revealing. A review projected that a major diagnostic error would be found in roughly 8% to 24% of autopsies at a modern US hospital. Yet autopsy has declined: in the UK in 2013 only 0.7% of adult inpatient deaths were followed by a consented autopsy. And a biopsy has a weakness of its own: it looks at a small piece. A cancer that is patchy can be missed if the needle goes through healthy tissue next to it. Interactive · Can a needle miss? Share of the organ with cancer Number of samples Assumes each sample independently lands in cancer with probability f. An invented illustration, not the performance of any real test; real biopsies are guided by imaging. If a fifth of the organ is involved, three samples all miss about half the time (51%); twelve miss under 7% of the time. That is why a "negative" biopsy is read with its limits in mind. What a graduate just sees Injury has a threshold: reversible, then not. Time is the variable. Necrosis is accident and apoptosis is design; both are normal words in the clinic. Redness, heat, swelling and pain are defence, and chronic inflammation is a fire that never went out. Tissues either regenerate or scar, and scar is a patch that never matches the original. A clot's damage depends on where it lodges and what backup the tissue has. A cancer is a clone, grown by exponential doubling, and is silent for most of its life. Metastasis, not the first lump, is usually what kills. A biopsy samples a small piece of a large, patchy organ, so a negative one is evidence, not proof. Where it shows up next Immunology (35/57) goes deep into the defence cells that inflammation calls up, and into allergy and autoimmunity. Pharmacology (35/56): many drugs work by damping inflammation, preventing clots or attacking dividing cells. Clinical Reasoning & Diagnosis (35/59): a diagnosis is a guess about which of these mechanisms is happening. In the world: pathologists who read the slides and sign the reports, cancer screening, wound care, and every news story about a "new" cause of cancer or heart disease. If you want to go deeper 📘 Textbook. Vinay Kumar, Abul Abbas and Jon Aster, Robbins & Kumar Basic Pathology (Elsevier). The standard first course; the larger Robbins and Kumar Pathologic Basis of Disease is the full reference. 📘 Textbook. Ivan Damjanov, Pathology for the Health Professions (Elsevier). A shorter text for nursing and allied-health programmes. 📘 Gentler. Siddhartha Mukherjee, The Emperor of All Maladies: A Biography of Cancer (2010). A history of cancer that reads like a story. 🌐 Free course. The US National Cancer Institute's What Is Cancer? and Khan Academy: Health and medicine videos. The takeaway Disease runs on a few repeating patterns: cells cross a threshold from hurt to dead, inflammation defends and sometimes won't stop, tissue regenerates or scars, blocked blood kills what lies downstream, and a clone of cells grows exponentially and sometimes spreads. Learn the patterns and each new disease becomes a variation on a theme. Try this: next time you get a bruise or a scrape, watch it daily. Note the order: redness and swelling first, then colour change, then fading. You are watching acute inflammation, clean-up and repair on a schedule. (If a wound is deep, red and spreading, or you are worried, ask a clinician.) Sources & Further Reading Sources. The mechanisms (reversible and irreversible injury, necrosis, inflammation, repair, thrombosis, neoplasia) are standard textbook pathology (Robbins). Apoptosis (Kerr, Wyllie and Currie, 1972; 2002 Nobel), Virchow's triad attribution, Celsus's four signs, Knudson 1971, Hanahan and Weinberg 2000, Armitage and Doll 1954, Morgagni 1761, Virchow 1858, wound-strength figures, reperfusion injury and the autopsy statistics were checked against encyclopaedia articles and their sources. Global cancer figures: IARC press release on GLOBOCAN 2022 (iarc.who.int). The tolerance times, immune-cell curves and backup fractions are illustrative; the doubling arithmetic, the tree and the biopsy probabilities were computed in node. Not medical advice. 🌐 NCI: What Is Cancer? 🌐 IARC Global Cancer Observatory 📘 Kumar, Abbas and Aster, Robbins & Kumar Basic Pathology 📘 Mukherjee, The Emperor of All Maladies Step 7. Immunology: An Army That Learns What to Fight and What to Spare ----------------------------------------------------------------------- Why: The immune system, from infection to allergy and autoimmunity. A few genes become billions of receptors, the enemy picks the cells that fight it, and the whole thing has to be stopped from hitting you. This is how defence, memory, allergy and cancer drugs fit together. A splinter goes in on Monday. By Tuesday the skin around it is red, warm and a little swollen, and by the weekend it is fine. You did nothing, but an army fought. Months later, a nurse gives you an injection of something that cannot make you ill, and you feel a sore arm for a day. Years after that, you meet the real germ and never notice. Somewhere in between, your body learned a face and kept the memory. Immunology is the study of that army: how it tells friend from foe, how it learns, why it sometimes fires at pollen or at its own joints, and how medicine now borrows it to fight cancer. This explains how medicine works and is not medical advice. For a health question, see a clinician; in an emergency, call your local emergency number. The course in one box Who takes itMedical, nursing, biomedical, pharmacy and biology students, usually in year 2. The medical-school version is organised around diseases; the biology version around cells and molecules. What it is forUnderstanding defence, memory and self-tolerance well enough to follow infection, vaccines, allergy, autoimmunity, transplants and cancer treatment. Comes afterCell Biology (35/44) and Microbiology (35/50). Leads toClinical Reasoning & Diagnosis (35/59) and Epidemiology & Evidence-Based Medicine (35/58) in the same major; Pathology (35/55) and Pharmacology (35/56) lean on it for inflammation and for biologic drugs. What this piece gives youYou will not be able to read a blood panel or choose a treatment. You will see how a few genes become billions of receptors, why the second encounter is faster than the first, why boosters exist, how tolerance can fail in two directions, and why blocking a brake can unleash an attack on a tumour. Easier first reads: Your Immune System, Day by Day (01/10) and Microbiology: The Invisible Majority (35/50), which already covers the vaccination threshold; here we look at what the vaccine does inside you. Two layers: fast and general, slow and specific Your body runs two defence systems. The innate system is old (animals, plants and even bacteria have versions) and works within minutes. Skin and mucus are barriers. Cells such as neutrophils and macrophages carry receptors that recognise patterns no healthy body part has: bacterial wall sugars, the double-stranded RNA of many viruses. When they sense one they swallow the intruder and release chemical alarms. Those alarms cause the red, hot, swollen skin: blood vessels widen and leak, and more cells flood in. Innate defence is quick but has no memory, and it responds about the same way each time. The adaptive system is the invention of jawed vertebrates. It takes days to start, but it is exquisitely specific and it remembers. Its two main players are B cells, which make antibodies (Y-shaped proteins that stick to a target), and T cells, which come in killer and helper forms and recognise fragments of proteins displayed on the surface of other cells. The two layers talk: innate cells that have swallowed a germ carry pieces of it to T cells, and without that introduction the adaptive system stays asleep. Whatever an antibody or T cell recognises is called an antigen. The big question of the subject is how a body that has never met a given germ can possibly have a cell ready to recognise it. Recognition by lottery: how a few genes make billions of receptors You cannot have one gene per possible germ; there are not enough genes, and new germs keep evolving. The answer, worked out by Susumu Tonegawa in the 1970s (Nobel Prize 1987), is a lottery run inside each developing cell. The antibody gene is built from pieces: a heavy chain from one V, one D and one J segment chosen from a menu, a light chain from V and J, and then the joints between the pieces are sloppily trimmed and padded with random letters. Every B cell shuffles its own deck once, and keeps the result for life. T cell receptors are made by the same trick. diversity ≈ (V × D × J) × (light-chain combinations) × junction variantsnumbers multiply, they do not add Interactive · A receptor lottery V segments for the heavy chain D segments J segments Light-chain combinations Junction variants (power of ten) Round, illustrative segment counts (real counts are of this order but differ by person, gene and species, and not every combination is functional). A body has about 3×1013 human cells (the 2016 estimate quoted in 35/50). With 40 × 25 × 6 = 6,000 heavy-chain combinations and 150 light-chain ones, the segments alone give 900,000 receptors. Add a thousand junction variants and it is 9×108. Push the junction slider to 9 and the possible repertoire is about 9×1014, some thirty times the cells in a body, so each of us carries only a sample. Most of the sample will never meet its target; the system works because it is huge, cheap and held in reserve. Clonal selection: the antigen picks its own army Having a million random receptors raises a new problem: how do you use the right one? Frank Macfarlane Burnet's answer, proposed in 1957, is clonal selection. Each B cell carries one receptor. When a germ's antigen fits it, that one cell is triggered, and with help from a T cell it multiplies into a clone. Some clone members become plasma cells, antibody factories; others become memory cells. The antigen does not instruct the cell; it picks from what was already there. It is natural selection, run inside you in days. Then it improves. In lymph-node structures called germinal centres, dividing B cells copy their antibody gene at a high error rate (somatic hypermutation). Most mutations make the antibody worse, a few make it grip tighter. Cells that grip better capture more antigen and get more help, so they win. The process is affinity maturation, and it explains why antibodies late in an immune response bind far more tightly than those made early. fraction bound = [antigen] ÷ ([antigen] + Kd)Kd is the concentration at which half of the antibody is occupied · lower Kd = tighter grip Interactive · Affinity maturation Antigen available (power of ten, arbitrary units) Rounds of selection A toy: 11 affinity classes with Kd from 1,000 down to about 1 (arbitrary units); each round, cells survive in proportion to the fraction bound; then 30% of cells mutate, of which 15% get tighter, 45% looser and 40% stay the same (invented split, but mostly-harmful is realistic). Expected values, no chance. Start with weak binders (mean Kd about 707). After 8 rounds with little antigen around (the default), selection is strict and the mean Kd falls to about 32, more than twenty-fold tighter. With a flood of antigen (103) nearly everyone binds and survives, so mean Kd only reaches about 363. Selection needs scarcity, which is why a response that clears a germ keeps sharpening for a while and then stops. Memory, vaccines and why boosters work The first meeting with an antigen is slow: about a week before antibodies are measurable, a peak after one to three weeks, then decline. Memory B and T cells change the second meeting. There are many more cells already tuned to the target, they need less coaxing, and the best of them have already been through affinity maturation. So the antibody level rises in days instead of weeks and climbs far higher. A vaccine is a way to stage the first meeting safely, with a harmless version or fragment of a germ (or the instructions for one), so that the real one is the second. Edward Jenner's cowpox experiment of 1796 was the first deliberate version of this idea (see 35/50 for what it means for populations). Why then is a booster ever needed? Two reasons, both visible in the model below. First, antibody levels from a single exposure fall over months as short-lived plasma cells die; a small pool of long-lived plasma cells and the memory cells remain, but the level may sink below what blocks infection quickly. A second exposure restarts the same rise with a bigger starting stock, and each round through the germinal centre improves affinity again. Second, some germs change (influenza drifts every year, which is why that vaccine is updated) so a new shot also teaches the army a new face. Waiting longer between exposures tends to give a stronger boost, because the response has time to mature: that is why vaccine schedules specify minimum gaps. Interactive · Antibodies after one, two or three exposures Number of exposures Days between exposures Antibody half-life (days) A teaching toy in arbitrary units, log scale. Peaks of 3, 24 and 60; lags of 7, 3 and 3 days; part of each response persists as a plateau; the dashed line at 2 is an invented "protective" level. Boost size shrinks if the gap is under six weeks. Not a vaccine schedule and nothing here is advice about any vaccine. With one exposure the level crosses the invented threshold of 2 for only about 35 of 730 days. With two exposures 60 days apart the second rise is faster (3-day lag instead of 7), peaks at about 25 (eight times the first) and stays above threshold for about 705 days. Shorten the gap to 14 days and the peak is lower (about 17); lengthen it to 180 and the first response has decayed, but the second peak is still about 25. That is the heart of every real schedule: a prime, a gap long enough for memory to form, then a boost. Telling self from not-self: tolerance, autoimmunity and transplants A lottery that makes random receptors also makes receptors that fit your own proteins. The body runs two safeguards. In the thymus, developing T cells are tested against self-fragments: those that bind too weakly to work are discarded, and those that bind self too strongly are mostly deleted (central tolerance). Out in the body, regulatory T cells and other checks restrain leakers (peripheral tolerance). The great majority of cells that start in the thymus never leave it. Burnet and Peter Medawar shared the 1960 Nobel Prize for the discovery of acquired immunological tolerance. Both safeguards trade one error for another. Delete too much of the repertoire and you have fewer cells to recognise foreign things; delete too little and some cells that attack self get out. When they do, the result is autoimmune disease: the immune system attacking the body's own tissue. Type 1 diabetes (insulin-making cells), multiple sclerosis (the insulation of nerves), rheumatoid arthritis (joint lining) and lupus (many tissues) are examples. Causes involve inherited risk, environmental triggers and chance together, are only partly understood, and many of these diseases are more common in women. Treatment is not self-treatment and is not covered here. Tolerance also explains transplant rejection. T cells read the surface proteins called HLA (human leucocyte antigens, the human version of the MHC) to know what is self. A donor organ with different HLA looks foreign, and a large share of the recipient's T cells react to it. The first successful organ transplant, a kidney between identical twins by Joseph Murray in Boston in 1954 (Nobel Prize 1990, shared with E. Donnall Thomas), avoided that problem; for everyone else, tissue matching and drugs that suppress the immune system (at the price of more infections) made transplants routine. The immune system does not know it is being unkind to a kidney that is saving a life. Interactive · Where to set the thymus threshold Deletion threshold (how strongly self-reactive before deletion) Share of escaping dangerous cells caught by regulatory T cells A cartoon: self-reactivity is a bell curve (mean 0, spread 1, arbitrary units); cells above the threshold are deleted; cells above 2 are "dangerous" if they escape. Real thymic selection is far harsher and more complicated. Illustrative, not data. At a threshold of 1.5, about 6.7% of the toy repertoire is deleted and no dangerous cell escapes. Raise it to 2.5 and only 0.6% is deleted, but 1.65% of cells are dangerous escapees, and with half caught by regulatory cells 0.83% still get out. Lower it to 0.5 and nearly a third of the repertoire is thrown away. No setting is free, which is the honest picture of tolerance: there is no threshold at which autoimmunity and weak defence both vanish. Allergy: the right machinery, the wrong target An allergy is an immune response to something harmless, such as pollen, a food protein or a venom. The common type is built on an antibody class called IgE. In a sensitised person IgE sticks to mast cells in skin, airways and gut; when the allergen binds two neighbouring IgE molecules the mast cell releases histamine and other chemicals within minutes: itch, swelling, sneezing, wheeze. Rarely the response is body-wide and dangerous (anaphylaxis), which is an emergency: call the local emergency number. IgE evolved, in the usual account, for defence against parasites such as worms; in places with few of them it is aimed at peanuts and pollen. Why allergy has become more common in many countries is still debated. What changed practice is a trial. The old advice was to keep babies away from peanuts. In the LEAP trial (Du Toit and colleagues, New England Journal of Medicine, 2015) 640 infants at high risk because of severe eczema or egg allergy were randomised to eat peanut or avoid it until age five. Among those with a negative skin test at the start, 13.7% of the avoidance group were allergic at 60 months against 1.9% of the peanut group; among those with a positive test, 35.3% against 10.6%. Early exposure helped, which fits the idea that the gut teaches tolerance. That advice depends on the child and a clinician, not on a general experiment at home. Immunotherapy: taking the brakes off Cancer cells are your own cells, so they are mostly tolerated. But some carry changed proteins that T cells can see, and tumours use the same brakes that protect healthy tissue. A T cell has built-in switches: signal 1 (the target), signal 2 (a "go" confirmation) and checkpoints, such as CTLA-4 and PD-1, that dampen an attack. Tumours can display PD-L1, which presses PD-1 and tells a T cell to stand down. Checkpoint inhibitors are antibody drugs that block the brake. James Allison (CTLA-4) and Tasuku Honjo (PD-1) shared the 2018 Nobel Prize in Physiology or Medicine; the first one, ipilimumab, was approved by the US FDA for melanoma in 2011. Older roots: William Coley injected bacteria into tumours in the 1890s, with erratic results. A second route, CAR-T cells, takes a patient's own T cells, equips them with an engineered receptor and returns them; the first was approved in 2017. The same brake guards healthy tissue, so removing it can cause autoimmune-type side effects (inflammation of the bowel, thyroid or lungs are well known). The tumour gains; some healthy tissue is also at risk. Interactive · Release the brake Tumour target signal Tumour PD-L1 (the brake) Checkpoint blockade (%) A cartoon: net signal = target signal + 2 (go signal) − 0.8 × PD-L1 × (1 − blockade). Attack when net is above 3. Healthy tissue: weak target signal 1.5 and PD-L1 3. All numbers invented to show the logic, not real patients or doses. With a target signal of 4 and a brake of 7, the tumour's net is 0.4, far under the threshold of 3: the T cell stands down. With full blockade it is 6.0. At 50% blockade it is 3.2, just over, while healthy tissue is at 2.3, still below. Pushed to 80%, healthy tissue creeps to 3.02. A real drug has a window, and no single threshold fits every patient or every tumour, which is why some cancers respond well and others hardly at all. What a graduate just sees There are two layers: a quick, general one that sets off alarms, and a slow, specific one that remembers. Receptors are shuffled from pieces; diversity comes from multiplication. The antigen does not teach the cell; it selects a cell that was already there. Selection sharpens antibodies, and it works best when antigen is scarce. Memory means the second meeting is faster and larger: this is what a vaccine buys. A booster restarts the rise from a bigger stock and improves the grip again; some boosters also update the target. Tolerance trades weak defence against self-attack; no threshold removes both. Allergy is a defence pointed at a harmless target; early exposure to food can lower risk in some children. Checkpoints are brakes. Taking them off helps against tumours and can also hurt healthy tissue. Where it shows up next Pathology (35/55): inflammation is the innate system's signature, and chronic inflammation links to many diseases. Pharmacology (35/56): monoclonal antibodies, checkpoint drugs and immunosuppressants are now a large share of new medicines. Epidemiology & Evidence-Based Medicine (35/58): vaccines and the peanut trial are the kind of evidence that piece teaches you to weigh. Bioinformatics & Genomics (35/52) and Genetics (35/46): HLA type shapes transplant matching and the risk of many autoimmune conditions. In the world: vaccination programmes, allergy clinics, transplant units, rheumatology and oncology. If you want to go deeper 📘 Textbook. Abul Abbas, Andrew Lichtman and Shiv Pillai, Cellular and Molecular Immunology (Elsevier). Widely used in medical schools; thorough and clinical. 📘 Textbook. Kenneth Murphy and Casey Weaver, Janeway's Immunobiology (W. W. Norton). The standard for the cell-and-molecule route, with good figures. 📘 Gentler. Matt Richtel, An Elegant Defense (2019): readable stories of immunologists and patients. Check any claim in it against a textbook. 🌐 Free course. The free OpenStax Microbiology textbook has solid immunology chapters, and Khan Academy's health and medicine courses cover the immune system, free. The takeaway The immune system is a learning army built on a lottery and a filter: random receptors, selection by the enemy, memory afterwards, and a set of brakes that stop it hitting you. Vaccines borrow the memory, allergy and autoimmunity are the filter misfiring, and modern cancer drugs borrow the brakes. Try this: next time a cut turns red and warm, name what you are seeing: innate alarm signals, not infection by itself. Check your own vaccination record with your clinic or health service; do not decide about shots on a web page. Sources & Further Reading Sources. Clonal selection (Burnet 1957, Nobel 1960 with Medawar), V(D)J recombination (Tonegawa, Nobel 1987), HLA, tolerance, IgE and mast cells are standard immunology (Abbas; Janeway). Murray's 1954 kidney transplant and the 1990 Nobel, and Jenner's 1796 experiment, are standard history. Checkpoint inhibitors: Nobel Prize 2018; ipilimumab FDA 2011; first CAR-T FDA 2017 (Wikipedia, Cancer immunotherapy). LEAP: Du Toit et al., N Engl J Med 2015;372:803–813. All interactive numbers were computed in node. This describes how medicine works; it is not medical advice. 📘 Abbas, Lichtman, Pillai, Cellular and Molecular Immunology 📘 Murphy and Weaver, Janeway's Immunobiology 🌐 OpenStax Microbiology (free) 📘 Richtel, An Elegant Defense Step 8. Microbiology: The Invisible Majority -------------------------------------------- Why: Bacteria, viruses and resistance. Most of the life on Earth, and most of what has ever lived on you, is too small to see. Microbiology learns how fast it grows, what it does, and how we have learned to live beside it, use it and fight it. A teaspoon of garden soil holds more bacterial cells than there are people on Earth. The bread you ate this week was raised by a fungus. The yogurt, the cheese, the pickle and the soy sauce were made by microbes, and so was the insulin in many a pharmacy fridge. And a cut that goes red, a cold that goes round a classroom, a hospital infection that no antibiotic touches: also microbes. Microbiology is the study of all of these: bacteria, archaea, viruses, fungi and a few other tiny things. It is where biology gets quantitative fast, because microbes multiply in minutes. This piece covers concepts only (what the groups are, how populations grow, how drugs and vaccines work in principle), not how to grow anything. It is not medical advice; if you are unwell, or worried about an infection, see a health professional. The course in one box Who takes itBiology, biomedical, nursing, food science, pharmacy, veterinary and medical students, usually in year 2 or 3, often with a lab. Medical schools teach a clinical version. What it is forKnowing the microbes that cause disease, keep us well and run the planet's chemistry, and the principles for living with them. Comes afterMolecular Biology (35/45), because most of the course is about microbial genes and cell machinery. Leads toGenetics (35/46) and Evolution (35/47) use microbes as the test case, Ecology (35/48) uses microbial communities, and a Medicine major goes on to immunology and infectious disease. What this piece gives youYou will not be able to identify a bacterium or treat an infection. You will see the size of the microbial world, why growth is exponential, what the microbiome evidence does and does not show, why resistance is evolution in a flask, and what a vaccination threshold means. Easier first reads: The Science of Bread (14/07), Fermentation: Letting Microbes Cook (14/08) and Your Immune System, Day by Day (01/10). Three domains and the viruses beside them Antonie van Leeuwenhoek looked at pond water through hand-ground lenses in the 1670s and saw "animalcules". Two centuries later Pasteur and Koch tied particular microbes to particular diseases. For a long time everything without a nucleus was called a bacterium. In 1977 Carl Woese and George Fox used ribosomal RNA to show that a group of odd microbes from swamps, hot springs and salt lakes were not bacteria at all. The living world has three domains: Bacteria, Archaea (single cells that look like bacteria but are in some of their molecular machinery nearer to us), and Eukarya (cells with a nucleus: animals, plants, fungi and protists). Fungi are eukaryotes, closer to animals than to plants: yeasts, moulds and mushrooms, with cell walls of chitin that digest food outside the body. Viruses are a different kind of thing: a strand of DNA or RNA in a protein coat that cannot copy itself without a host cell's machinery. Whether to call them alive is a definition argument, not a measurement; what is clear is that they are enormously abundant, and the most numerous kind, the bacteriophages that infect bacteria, are estimated at around 1031 particles on Earth. A quick, non-clinical way to sort bacteria is the Gram stain, invented by Hans Christian Gram in 1884: a thick wall of peptidoglycan keeps the purple dye (Gram-positive), a thin one lets it wash out (Gram-negative, which also have an outer membrane). It matters later, because it shapes which drugs reach which cells. Interactive · Zoom into the invisible Field of view (log scale) Typical diameters (they vary a lot): hair 70 µm, skin cell 30 µm, yeast 8 µm, E. coli about 2 µm long, a typical virus 0.1 µm (100 nm). Circles are drawn at true relative size; anything under one pixel is shown as a dot. Set the field to 1 mm (1,000 µm) and a human hair is a 7% sliver of the view while a bacterium would need 500 side-by-side to cross it. Set it to a hair's width and you can fit 35 E. coli end to end. A typical virus is about twenty times smaller than the bacterium it can infect, which is why ordinary light microscopes cannot see one. Growth curves: doubling is the whole game A bacterium grows, then divides in two. If conditions are good, each cell of a fast species such as Escherichia coli can divide about every 20 minutes. That is three doublings an hour, and 230 after ten hours: one cell becomes more than a billion (1.07 billion). Left unchecked for two days, the mass would exceed the Earth's. Nature never allows that, because food runs out, waste builds up and the culture stops. N(t) = N0 × 2t / tdcells now = cells at the start × 2 to the power (time ÷ doubling time) · growth stops near the carrying capacity A culture's life in a closed container has four phases. A lag phase while cells adjust and make the right enzymes; an exponential phase, a straight line on a log plot; a stationary phase, as food runs out; and a death phase. Biologists therefore plot on a log scale: constant doubling then looks like a straight line, and the slope is the growth rate. What Ecology (35/48) says about logistic growth holds here too, since a flask is a population with a ceiling. Interactive · A culture, as a model Doubling time (minutes)Lag phase (hours) Model, not an experiment: start at 1,000 cells/mL, capacity 109 cells/mL (illustrative), N = 1 / (1/Nexp + 1/K) with Nexp = N0·2(t − lag)/td. Not a protocol. With a 30-minute doubling time and a one-hour lag, the model reaches half of its capacity at about 11 hours and is flat soon after. Slow the doubling to 3 hours and in 24 hours it has grown only about 200-fold, nowhere near the ceiling. A small change in doubling time is a huge change in numbers: that is why food left warm goes off so much faster than food left cold. The microbiome without the hype Your gut, skin and mouth are inhabited by communities of microbes, most of them harmless and many of them useful. The old claim that microbes outnumber your own cells ten to one has been revised: a 2016 estimate by Ron Sender, Shai Fuchs and Ron Milo puts it at roughly one to one, about 38 trillion bacteria and 30 trillion human cells in a typical adult, with large uncertainty. The genes are a different story: the microbial genes in a gut outnumber human genes by something like a hundred to one. What is well supported: gut microbes help break down fibre and make some vitamins and short-chain fatty acids; they train the immune system, especially early in life; and a stable community resists invaders. The clearest medical use is also the most dramatic: for repeated Clostridioides difficile infections, which can follow antibiotics, restoring a healthy community has worked well enough that microbiome-based treatments have been approved by regulators in recent years. What is less well supported are the sweeping claims: that your microbiome "causes" obesity, mood or autism in people, that a probiotic yogurt will "rebalance" a healthy gut, or that a single stool test can score your health. Many links come from mouse studies or correlations in small human samples, and cause and effect are hard to separate. A community's diversity is the idea that holds up best, and ecologists already have a number for it. H = − Σ pi ln piShannon diversity: p = share of each species · higher when there are more species and when they are more even Interactive · Counting a community EvenFour, unevenOne dominatesSpecies A (relative abundance)Species BSpecies CSpecies D Invented abundances for illustration, not data from any real gut. Shannon H uses natural logs; evenness J = H / ln(richness); Simpson index = 1 − Σp². Four species at 40, 30, 20 and 10% give H = 1.28; four equal species give the maximum, ln 4 = 1.39; a community that is 97% one species is about 0.17. The same number can be raised by adding species or by evening them out. Antibiotics tend to reduce diversity in the gut, which usually recovers, though not always fully; and the question of which diversity matters for health is open. Antibiotics and resistance: evolution in a flask An antibiotic is a poison chosen to hit something bacteria have and we do not. Penicillin, found by Alexander Fleming in 1928 and made into a medicine in the 1940s by Howard Florey, Ernst Chain and colleagues, blocks the building of the bacterial cell wall, a structure human cells lack. Other classes target the bacterial ribosome (which differs from ours), the enzymes that copy bacterial DNA, or the pathway for making folate. They do nothing against viruses, which have none of these targets. Resistance is natural selection in plain view. In 1943 Salvador Luria and Max Delbrück showed that bacterial resistance mutations arise at random before the drug is added, not in response to it. Among a billion cells there are a few that happen to survive; the drug removes their competitors; they take over. Bacteria can also swap resistance genes between species on small DNA circles called plasmids. A 2022 analysis in The Lancet estimated that in 2019 resistant bacterial infections were directly responsible for about 1.27 million deaths worldwide and associated with 4.95 million. f′ = R / (R + S × (1 − kill))after one round of drug: new resistant share = resistant ÷ (resistant + surviving susceptible) · the population then regrows Interactive · Selection, round by round Resistant fraction before the drug (log₁₀)Drug kill of susceptible cells (log₁₀)Rounds of drug Expected values for 109 cells that regrow to the same size between rounds, with no cost of resistance. Illustrative: a real infection also has the immune system, drug levels that vary, and chance. Not guidance on treatment. With one resistant cell in ten million (10−7) among a billion, 100 cells are resistant at the start. A drug that kills 99.9% of the susceptible cells leaves their share at 10−4 after round one, about 0.09 after round two and 99% after round three. A stronger kill (more nines) hastens the takeover of the survivors, since competitors are removed. That is why microbiologists talk about slowing resistance and not stopping it: stewardship means using drugs only where they will work, and matching the drug to the target, and why you take a prescription as your clinician directs. Viruses and vaccines: the threshold idea A virus is at heart a copy instruction. It enters a cell, hijacks its machinery, and the cell makes thousands of new particles. For public health the key number is the basic reproduction number R0: how many people one case infects in a fully susceptible population. If it is above 1, an outbreak grows; below 1, it fades. R0 is not a fixed constant of a virus: it depends on contacts, climate and behaviour. Measles is commonly quoted at 12 to 18, among the highest known; seasonal influenza is far lower, nearer 1 to 2. A vaccine, in concept, shows the immune system a harmless version or piece of a pathogen (a weakened or inactivated virus, a purified protein, or the instructions to make one) so that memory cells are ready when the real one arrives. Edward Jenner's 1796 experiment with cowpox began it; smallpox was declared eradicated in 1980, and rinderpest, a cattle virus, in 2011. If each case infects R0 people but a fraction of contacts are protected, the effective number falls: Reff = R0 × (1 − coverage × effectiveness)  ·  threshold = (1 − 1/R0) ÷ effectivenessoutbreaks fade when Reff < 1 · "herd" protection is when enough are immune that chains break Interactive · How much is enough? Flu-like (1.5)3Measles-like (15)Basic reproduction number R₀Vaccine effectiveness (%)Coverage (%) R₀ values are illustrative and vary by place and time. Simple well-mixed model: no waning immunity, no clustering of the unvaccinated, which in real life lowers protection. Describes how the idea works; not advice about any vaccine. At R0 = 3 and a vaccine that is 90% effective, you need 74% coverage (0.667 ÷ 0.9). At 15 you need 1 − 1/15 = 93.3% immune, so about 104% coverage with a 90% vaccine: not reachable, which is why a measles-like virus returns when coverage slips even a few points, and why immunity from earlier infection or a second dose counts. Real populations are lumpy: unvaccinated people cluster in families and schools, so the average can hide pockets that sustain outbreaks. Microbes at work: bread, soil, medicine and industry Most microbes do chemistry we depend on. Yeasts ferment sugars to carbon dioxide and alcohol: the bubbles in bread, the base of beer and wine. The East Asian mould Aspergillus oryzae (koji) makes soy sauce, miso and sake; lactic acid bacteria sour yogurt, kimchi, sauerkraut and Ethiopia's injera batter; bacteria raise cassava gari in West Africa. In soil, Rhizobium bacteria in the roots of beans, peas and clover turn nitrogen from the air into a form plants can use, and fungi called mycorrhizae trade nutrients with roots (the popular "wood wide web" story of trees talking through them has been criticised in a 2023 review as going beyond the evidence). Sewage plants are managed microbial communities. Industry uses them as factories. In 1978 a team at Genentech and City of Hope made human insulin in engineered E. coli, and human insulin made that way reached patients in 1982. The heat-stable DNA-copying enzyme in PCR, Taq polymerase, comes from Thermus aquaticus, a microbe from a Yellowstone hot spring. Fungi cause trouble too: a chytrid fungus has driven declines in hundreds of amphibian species worldwide, which links the subject back to ecology. Interactive · A rising loaf: how much sugar? Dough volume (mL)Rise wanted (× original volume)Share of the gas the dough holds (%) Ideal gas at 25 °C and 1 atm (24.46 L per mole). Yeast fermentation C₆H₁₂O₆ → 2 C₂H₅OH + 2 CO₂ (glucose 180.16 g/mol). Real dough also dissolves some CO₂, and flour supplies sugar from its starch. A back-of-envelope, not a recipe. To double a 500 mL lump of dough, you need 500 mL of new gas, which is 0.0204 mole of CO2 and so 0.0102 mole of glucose, about 1.8 g of sugar, if the dough holds every bubble. If the dough holds only 60% of the gas, the figure is 3.1 g. A few grams of sugar will raise a loaf; flour starch, broken down by enzymes, supplies it. Fermentation is a few lines of chemistry run by millions of cells. What a graduate just sees The microbial world is the majority of living things; animals and plants are a small branch. Bacteria, archaea and eukaryotes are three domains; viruses sit outside the tree, dependent on a host. Growth is a doubling; a small change in doubling time is a huge change in numbers. Log scales turn exponential growth into a straight line. The microbiome is real and important; most dramatic health claims about it are ahead of the evidence. Antibiotics aim at structures bacteria have and we do not, which is why they do nothing to viruses. Resistance is selection: the variants are there first, and the drug picks them. Vaccination works by pushing the reproduction number below 1; the threshold rises steeply with R0. Many foods, medicines and materials are by-products of microbial chemistry. Where it shows up next Genetics (35/46) and Molecular Biology (35/45): bacteria and phages are where most of the molecular toolkit was found, from restriction enzymes to Taq. Evolution (35/47): a flask of bacteria is evolution you can watch in days. Ecology (35/48): soil and ocean microbes run the carbon and nitrogen cycles. Physiology (35/49) and, in a medicine major, immunology and infectious disease: fever, inflammation and the immune response are the body's side of the story. In the world: hospital infection control, food safety, vaccination programmes, brewing and dairies, water treatment, biotech, and antimicrobial stewardship. If you want to go deeper 📘 Textbook. Michael Madigan and colleagues, Brock Biology of Microorganisms (Pearson). The standard university text, the one MIT's microbiology course lists. 📘 Textbook. Joanne Willey, Kathleen Sandman and Dorothy Wood, Prescott's Microbiology (McGraw Hill). The other widely assigned general text. 📘 Gentler. Ed Yong, I Contain Multitudes (2016). A readable, balanced tour of the microbiome. 🌐 Free course. MIT OpenCourseWare 20.106J Systems Microbiology (Fall 2006), and the free OpenStax Microbiology textbook. The takeaway Microbes are small, abundant and fast. They divide exponentially until something stops them, they swap genes, they are the first chemists of the planet, and they evolve in front of us. That one picture explains why infections grow, why drugs lose power, why vaccines have thresholds, and why bread rises. Try this: this week, find three foods in your kitchen that microbes made (bread, yogurt, cheese, vinegar, soy sauce, pickles). Ask which kind: yeast, bacterium or mould, and what it did to the food. (Observation only; do not taste anything that looks or smells off.) Sources & Further Reading Sources. Domains (Woese and Fox, 1977), the Gram stain (1884), growth phases, antibiotic targets, plasmids, R0 and the vaccination threshold are standard textbook microbiology (Madigan; Willey). The Luria–Delbrück experiment (1943), and the microbial cell counts of Sender, Fuchs and Milo (2016), were checked against encyclopaedia summaries. The antimicrobial-resistance burden is Murray et al., The Lancet (2022). Smallpox eradication (WHO, 1980), rinderpest (2011), recombinant insulin (1978, 1982) and Taq polymerase are standard history, given in outline. Not medical advice. All interactive numbers were computed in node. 🌐 MIT OCW 20.106J Systems Microbiology 🌐 OpenStax Microbiology (free) 📘 Madigan et al., Brock Biology of Microorganisms 📘 Yong, I Contain Multitudes Step 9. Pharmacology: What a Drug Does, and What the Body Does Back ------------------------------------------------------------------- Why: How drugs act on the body, and the body on drugs. A drug is a molecule that fits a lock, and a dose is a number of keys. Pharmacology is the course that learns how many keys do what, how long they stay in the body, and why the same dose is a cure for one person and a problem for another. A cup of coffee gives you a lift within half an hour. Two cups give a stronger one, and ten give you the shakes. Have one at night and you may still be awake at midnight, because the caffeine is still in your blood, and the person next to you, who drank the same, is asleep. Four ideas are hiding in that one cup: the effect depends on the dose; there is a ceiling and a downside; the drug leaves the body at a particular speed; and bodies differ. Those four ideas are pharmacology. Four hundred years before it had a name, the physician Paracelsus wrote that the dose, not the substance, decides whether something is a poison. This piece explains how medicine works and is not medical advice. It gives no doses of any real medicine: the numbers in the interactives are for a made-up "drug X" in made-up units. For a health question, see a clinician or pharmacist; in an emergency, call your local emergency number. The course in one box Who takes itMedicine, pharmacy, nursing, dentistry, veterinary and biomedical students, usually in the second year of medicine, and across several years in pharmacy. It splits into pharmacodynamics (what the drug does to the body) and pharmacokinetics (what the body does to the drug), then into specific drug classes. What it is forUnderstanding why drugs work, why they harm, and how to choose a drug and a schedule without guessing. Comes afterBiochemistry (35/43), because drugs act on proteins, and Physiology (35/49), because they act on loops. Leads toSurgery, Anaesthesia & Emergency Care (35/62), where drug action is minute-to-minute; and Clinical Reasoning & Diagnosis (35/59) and the clinical years. What this piece gives youYou will not be able to choose or dose a drug, and nothing here is a dosing guide. You will see the S-shaped dose–response curve, why one drug blocks another, what a therapeutic index is, how half-life sets a schedule, why bodies differ, and how a drug is tested before approval. Easier first reads: Physiology (35/49) and Biochemistry (35/43). Receptors and dose–response: more drug, more effect, then a ceiling Most drugs work by binding to a protein, usually a receptor (a lock on the surface of a cell that the body's own messengers, such as adrenaline, normally open). The idea that cells carry "receptive substances" for specific chemicals dates to the early 1900s. A drug that fits and switches the lock on is an agonist. The more drug there is, the more receptors are occupied, until they are all busy and more drug does nothing: a ceiling (the maximum effect, Emax). effect = Emax × Cn ÷ (EC50n + Cn)C = amount of drug · EC₅₀ = the dose giving half the maximum · n = steepness (n = 1 for simple one-to-one binding) On an ordinary scale this curve bends flat; on a log scale it becomes an S-shape, and that is how pharmacologists always draw it. Two numbers define a drug: potency (how far left the S sits, set by EC50) and efficacy (how high it goes). They are not the same: a potent drug is simply effective at a small dose, not a stronger one. Interactive · The S-curve Dose (log scale, X-units) Potency: dose for half effect Steepness (n) The occupancy-style (Hill) equation with a made-up drug X in made-up units. No real medicine is represented. For simple binding (n = 1) the dose must rise 81-fold to carry the effect from 10% to 90%, which is why a log axis is needed. A steeper drug (n = 2) covers that span in a 9-fold change, so a small error in dose matters more. Agonists and antagonists: keys, blanks and partial keys A molecule can fit the lock without turning it. An antagonist binds to the receptor and does nothing except block the way. This is a profound idea: much of modern medicine works by blocking. The Scottish pharmacologist James Black built propranolol, which blocks adrenaline's beta receptors on the heart, and cimetidine, which blocks histamine receptors in the stomach; he shared the 1988 Nobel Prize in Physiology or Medicine for this "rational" design. The best-known rescue example is naloxone, which sits on opioid receptors and displaces opioids such as morphine. There are four useful cases. A competitive antagonist fights the agonist for the same lock; flood the system with agonist and you can still win, so the curve slides right but the ceiling is unchanged. The shift is predictable: the dose ratio = 1 + [antagonist] ÷ its binding constant (the Schild relation, after Heinz Otto Schild). A non-competitive (insurmountable) blocker lowers the ceiling itself. A partial agonist turns the lock only part-way, so its ceiling is low, however much you give; it also stops a full agonist from working fully, which is exactly how buprenorphine is used with opioids. Interactive · Add a second drug NoneCompetitive blockerPartial agonistNon-competitive blocker Amount of second drug Standard textbook models (Schild shift; reduced ceiling; fixed 50% partial agonist). The partial agonist chip ignores the slider because its ceiling is fixed at 50%. Made-up drugs. With a competitive blocker at five times its binding constant (slider at 10), the agonist needs 6 times the dose for the same effect. At 20 times (slider at 40) it needs 21 times. The shift is the signature of competition. The therapeutic index: how far apart good and harm are Every effective drug is also harmful at a high enough dose, so what matters is the gap. Doctors speak of the therapeutic index: the dose that causes harm in half of people divided by the dose that works in half of people. A wide index (say, 100) gives room for error. A narrow one (say, 2 or 3) means the working dose and the harmful one overlap in the population, and treatment needs close monitoring, often with blood tests. Familiar narrow-index medicines include the blood thinner warfarin, the heart drug digoxin and lithium. They are not unsafe; they are respected. therapeutic index = TD50 ÷ ED50dose harmful in half the people ÷ dose effective in half the people Interactive · The gap between working and harming Therapeutic index Dose Two idealised population curves of equal steepness (n = 3), as a teaching picture. Real drugs have messier curves, and harm is often more than one effect. With an index of 10 and a dose three times the median effective dose, about 97% benefit and 3% are harmed. A stricter measure compares the dose that works in 99% with the dose that harms 1%; at an index of 10 those curves overlap (the ratio is 0.47), and it takes an index near 100 to open a gap (4.7). A narrow index is a design constraint on every prescription. Pharmacokinetics: what the body does to the drug A drug that works in a test tube still has to arrive. Pharmacokinetics tracks four steps, remembered as ADME: absorption (into the blood, often through the gut wall), distribution (out of the blood into tissues), metabolism (usually chemically changed by enzymes, mostly in the liver) and excretion (usually by the kidneys, into urine). A pill swallowed goes first through the liver on the way to the rest of the body, where part is destroyed: the "first-pass effect", one reason oral and injected versions of a drug need different amounts. The one quantity worth keeping is the half-life: the time for the level to fall by half. After one half-life 50% is left, after two 25%, and so on. Most drugs clear in proportion to what is there, so the level follows an exponential, and the same arithmetic gives the schedule. In terms of numbers: half-life = 0.693 × V ÷ CL, where V is the volume the drug seems to spread through and CL is clearance, the volume of blood cleared of the drug per hour. C(t) = C0 e−kt · k = CL ÷ V · t½ = 0.693 ÷ kone-compartment model after a single injection · repeated doses build up and level off after about 3 to 5 half-lives Here is a fictional drug X: a 400-unit dose into a 40-litre volume gives 10 units per litre at the start. Change the half-life and dosing and watch the curve. With a half-life of 8 hours and a dose every 12 hours, the highest level climbs from 10 to about 15.4 units/L as the doses overlap, an accumulation factor of 1.55; it takes about 27 hours (3.3 half-lives) to reach 90% of its final level. A swallowed dose rises more slowly, peaks lower (about 6.3 for a single dose, with 80% of it getting in) and falls with the same half-life. Interactive · Drug X in the blood Injected (all at once)Swallowed Half-life Dosing interval Number of doses Fictional drug X: 400 units per dose, 40 L volume, swallowed form absorbed with rate 1 per hour and 80% reaching the blood. Round numbers for practice. Not a model of any real medicine and not a dosing tool. Interactions and individual variation: same drug, different bodies Real people take more than one drug and differ in the enzymes that clear them. Some of the most important differences are in the cytochrome P450 liver enzymes. A slow metabolizer clears a drug slowly, so ordinary doses pile up; an ultrarapid one clears it fast, so the same dose may do too little. When the drug is a prodrug (it must be changed in the body to become active), the effect is reversed. Codeine, converted by the enzyme CYP2D6 to morphine, is the standard example, and regulators have restricted it in children partly because of fast converters. Frequencies differ by ancestry: reviews report that roughly 7% to 10% of people of European ancestry have reduced CYP2D6 function, with a smaller share in many other groups. Other kinds of variation are about the immune system, not clearance. An HLA gene variant, HLA-B*15:02, that is more common in some East and Southeast Asian ancestries, is linked to a rare, severe skin reaction to the epilepsy drug carbamazepine, so prescribing information advises considering a genetic test before it is started in people of those ancestries. Interactions work the same way: an inhibitor of an enzyme acts like a slow metabolizer, and an inducer like a fast one. Grapefruit juice blocks an enzyme in the gut wall and can raise levels of some blood-pressure drugs. Age, kidney and liver function, pregnancy and body size all move clearance as well. Interactive · Four people take the same dose TypicalSlow metabolizerUltrarapidTaking an inhibitor Drug X again: 400 units every 12 hours, volume 40 L, typical clearance 4 L/h. Clearance multipliers 1, 0.4, 2.5 and 0.5 are illustrative. The "useful window" is made up. Average level = dose ÷ (clearance × interval). Clearance changes from 4 to 1.6 L/h and the same schedule goes from 8.3 to 20.8 units/L, above the window; at 10 L/h it falls to 3.3, below it. Not every drug is proportional, either. When the enzymes are close to full, an extra dose is cleared no faster, and the level climbs out of proportion (alcohol and the epilepsy drug phenytoin are textbook cases). Interactive · When clearance saturates ProportionalSaturable Dose rate Michaelis–Menten clearance for drug X (maximum 40 units/h, half-saturation 5 units/L), compared with a proportional model that matches it at low levels. Made-up numbers. Double the dose rate from 10 to 20 units/h and the proportional model doubles the level (1.25 to 2.5). The saturable one triples it (1.7 to 5.0); at 30 it is 15, at 39 it is 195, and at 40 there is no steady level at all. From lab to approval: how a drug earns its place A candidate drug passes through preclinical work (cells, then animals) and then four phases in people. Phase 1: a few dozen volunteers, to see how it is handled and whether it is tolerated. Phase 2: a few hundred patients, to see whether it seems to work and at what dose. Phase 3: hundreds to thousands, usually randomised and often blinded, against a placebo or the current treatment. Then regulators (the US FDA, the European Medicines Agency, the UK's MHRA, Japan's PMDA, China's NMPA, Brazil's ANVISA, Mexico's COFEPRIS and others) weigh the evidence. Phase 4 and surveillance continue after approval, because rare harms only show up in large populations. The whole process often takes a decade or more. The system was built from disasters. In 1937 the US elixir sulfanilamide, a medicine dissolved in a poisonous solvent, killed about 107 people and led to the 1938 Food, Drug, and Cosmetic Act. Thalidomide, sold in dozens of countries as a sedative and for morning sickness from the late 1950s, caused serious birth defects in more than 10,000 children before Widukind Lenz linked it to them in 1961. In the US, the reviewer Frances Kelsey had held up its approval, and the 1962 Kefauver–Harris Amendment required that drugs be shown to be safe and effective. Every phase is, in effect, an answer to something that once went wrong. Most candidates fail. One analysis of drugs in development between 2003 and 2011 (Hay and colleagues, 2014) found that only about 10% of those entering clinical trials were approved. Failure is mostly for lack of benefit or unexpected harm, which is why a trial is a test and not a formality. Drag the survival rate of each stage below; these are round numbers close to that picture, not the paper's exact figures. Interactive · The attrition funnel Drugs entering human trials Pass phase 1 Pass phase 2 Pass phase 3 Regulator approves Illustrative round stage rates (65%, 30%, 58%, 90%) multiplying to about 10%. Real rates differ by disease area, era and company. At those defaults, 1,000 candidates become about 102 approvals. Notice where the loss is: the second phase, where the first honest test of benefit happens, takes the most. What a graduate just sees Effect is an S-curve on a log scale: potency is where it sits, efficacy is how high it goes. Blocking is as important as activating; a competitive blocker shifts the curve, a non-competitive one lowers the ceiling. Safety is the gap between the dose that works and the dose that harms; narrow gaps mean monitoring. Half-life sets the rhythm: about four half-lives to clear a dose or build to steady level. Prodrugs, inhibitors and genes mean the same pill is a different dose in different people. Clearance can saturate, and then small dose changes make big level changes. A drug's journey from lab to market is a funnel, and most of it fails for real reasons. The rules exist because of thalidomide and elixir sulfanilamide; every safeguard has a story. Where it shows up next Physiology (35/49): receptors and feedback loops are the targets; a drug is a push on a loop. Pathology (35/55): anti-inflammatory, anti-clotting and anti-cancer drugs each target a mechanism you have seen there. Surgery, Anaesthesia & Emergency Care (35/62): anaesthesia is controlled pharmacokinetics, in minutes. In the world: pharmacists and prescribers, generic-drug policy, the WHO's essential medicines idea, drug-label warnings, and every headline about a trial result. If you want to go deeper 📘 Textbook. Humphrey Rang, Maureen Dale and colleagues (Ritter, Flower, Henderson and others), Rang & Dale's Pharmacology (Elsevier). A widely used text, strong on mechanisms. 📘 Textbook. Bertram Katzung (ed.), Basic & Clinical Pharmacology (McGraw Hill). The common US medical-school reference. 📘 Gentler. Ben Goldacre, Bad Pharma (2012). A critical, readable account of how trials are done, hidden and reported. 🌐 Free course. The US FDA's explanation of drug development and the IUPHAR/BPS Guide to Pharmacology, a free database of receptors and drugs. The takeaway A drug is a key for a lock, and its effect is an S-curve with a ceiling. Others can block, partly turn or lower the ceiling; safety is the gap between good and harm; the body clears the drug at a rate that sets its schedule; people differ; and the whole thing is tested because most candidates fail. The right question about any medicine is never just "does it work?" but "how much, in whom, for how long, and with what gap before harm?" Try this: next time you read a medicine's information leaflet, find three things it tells you: what the drug does (the target), how often it is taken (the half-life), and which other drugs or foods to avoid (the interactions). Do not change anything; just read it as a pharmacologist would. Ask your pharmacist if anything is unclear. Sources & Further Reading Sources. The Hill/Emax model, Schild relation, therapeutic index, one-compartment pharmacokinetics and Michaelis–Menten clearance are standard textbook pharmacology (Rang & Dale; Katzung). James Black and the 1988 Nobel Prize, the Schild relation, CYP2D6 variation, the HLA-B*15:02 and carbamazepine link, grapefruit interactions, Paracelsus's dose principle, clinical-trial phases, elixir sulfanilamide (1937; 107 deaths; 1938 Act), thalidomide (46 countries; more than 10,000 children; Lenz 1961; Kelsey; Kefauver–Harris 1962) were checked against encyclopaedia articles and their sources. Hay et al., Nature Biotechnology 32:40–51 (2014) for the approximate 10% overall success rate (checked through a secondary summary; the individual stage rates in the interactive are round, illustrative numbers). Drug X is fictional; all interactive numbers were computed in node. Not medical advice. 🌐 FDA: The drug development process 🌐 IUPHAR/BPS Guide to Pharmacology 📘 Rang & Dale's Pharmacology 📘 Goldacre, Bad Pharma Step 10. Epidemiology & Evidence-Based Medicine: Out of How Many, Compared With What? ------------------------------------------------------------------------------------- Why: How medicine knows what works. A food that "raises cancer risk by 20%", a screening test with a scary false-alarm rate, a trial that overturned a medical belief: how medicine counts, compares and sometimes gets it wrong. The headline says that a food "raises your risk of cancer by 20%". Should you worry? It depends on a number the headline left out: 20% of what? If the usual risk is 5 in 100, a 20% rise is 6 in 100. If the usual risk is 1 in 10,000, it is 1.2 in 10,000, which is two people in a hundred thousand more. Both sentences are true. Only one of them should change your dinner. Epidemiology is the science of counting disease in populations and asking what causes it. Evidence-based medicine takes those methods to the bedside: before a treatment is used, ask who tested it, against what, and how big the effect really was. This explains how medicine works and is not medical advice. For a health question, see a clinician; in an emergency, call your local emergency number. The course in one box Who takes itMedical students (often as "epidemiology and biostatistics" or "evidence-based medicine", in years 2 and 3), plus public-health, nursing and pharmacy students. Programmes that follow the GMC and Canadian CanMEDS outcomes treat it as a core skill for every doctor, not a specialism. What it is forJudging whether a claim about health is likely to be true, and how much it matters for a person. Comes afterProbability & Statistics (34/05) and Research Methods in Psychology (35/19), which cover sampling, p-values and why random assignment matters. Leads toClinical Reasoning & Diagnosis (35/59) and Public & Global Health (35/61). What this piece gives youYou will not be able to design a trial or run a meta-analysis. You will see why the same effect looks big or small depending on how it is stated, why the obvious comparison can lie, what screening costs, and how a pile of trials becomes a conclusion. Easier first reads: Bayes' Theorem: What a Positive Test Really Means (12/17) and Statistics for Reading the World (12/05). Rates and risks: always ask "out of how many?" Four numbers carry most of public health. Prevalence is how many people have a condition now. Incidence is how many new cases appear in a period among people who could get it. Risk is the chance that one person gets the outcome over a stated time. A rate divides cases by person-time, so a person followed for two years counts twice as much as one followed for one. And odds are risk ÷ (1 − risk), the same information in a different form that behaves better in some statistics. Prevalence and incidence are linked: in a steady state, prevalence ≈ incidence × average duration. A condition that is rare to catch but lasts forty years, such as some chronic diseases, can be common in the population; a flu that is caught by many but lasts a week can be rare on any given day. Every comparison starts with a denominator. A city with more cancer deaths than a village may simply have more people, or more older people. The reason epidemiologists "age-standardise" death rates is to stop an old country looking sicker only because it is old. Study designs: from the case report to the randomised trial Designs form a ladder of how well they rule out other explanations. A case report tells one story. A case-control study starts from people with a disease and picks similar people without it, then looks back at what they were exposed to: cheap and quick for rare diseases. In 1950 Richard Doll and Austin Bradford Hill used it in the British Medical Journal to show that people with lung cancer were far more likely to have smoked. A cohort study follows groups with and without an exposure forward in time; their 1951-onward study of British doctors added the evidence that smokers died younger. A randomised controlled trial (RCT) assigns the exposure by chance, so groups differ only by luck, which makes it the strongest design for testing a treatment. The first published RCT in medicine is usually taken to be the Medical Research Council's trial of streptomycin for tuberculosis in 1948, with Hill as its architect. Each design gives a different number. A cohort gives a relative risk (RR: risk in the exposed ÷ risk in the unexposed). A case-control study cannot see risks, because the investigators chose how many cases and controls to include, so it gives an odds ratio (OR). For rare diseases the OR is almost the same as the RR. For common ones it overstates it: RR = r₁ ÷ r₀  ·  OR = [r₁ ÷ (1 − r₁)] ÷ [r₀ ÷ (1 − r₀)]r₀ is the risk without the exposure, r₁ with it · the OR is always further from 1 than the RR Interactive · Relative risk or odds ratio? Risk without the exposure (%) True relative risk Invented numbers for illustration. Risk with the exposure is capped at 99.9%. Number needed to treat or to harm = 1 ÷ absolute difference in risk. At a baseline of 1% and a true RR of 2, the OR is 2.02: close enough that nobody would notice. At a baseline of 30%, the same RR of 2 gives an OR of 3.5. When a newspaper reports an odds ratio as if it were "times more likely", common outcomes are exaggerated. Rare outcomes are not. Confounding and bias: why the obvious comparison lies A confounder is a third thing that is linked to both the exposure and the outcome, and creates a link between them that is not causal. Coffee drinkers get more lung cancer? In some old data, yes, because more coffee drinkers also smoked. Hormone therapy for menopausal women looked, in observational studies, as if it protected the heart, partly because women who chose it tended to be healthier and richer already (the "healthy user" effect). A randomised trial breaks the link by assigning the exposure by chance. Without one, the main defence is to compare like with like (stratify or adjust), which only works for confounders you thought to measure. Bias is different: a systematic error built into the way people were chosen or measured. Selection bias (who ended up in the study), recall bias (people with a disease remember their past differently), and lead-time bias (below) are three. Luck is a third source of error again, and shrinks as samples grow; bias and confounding do not. Interactive · A confounder that fakes an effect Coffee drinkers who smoke (%) Non-drinkers who smoke (%) Crude comparisonSplit by smoking An invented population. By construction, coffee does nothing: smokers have a 10% risk of the outcome and non-smokers 1%, whatever they drink. Illustrative, not real data. With 40% of drinkers and 15% of non-drinkers smoking, drinkers have a 4.6% risk against 2.35% for non-drinkers: a crude relative risk of about 1.96, from a drink that does nothing. Split into smokers and non-smokers, the risk ratio inside each group is exactly 1. Set both sliders equal and the false effect vanishes. That is what a graduate sees in every "X is linked to Y" headline: what else travels with X? Absolute and relative risk, and the number needed to treat A treatment that "halves the risk" sounds great, but what it does for you is the absolute risk reduction (ARR): the baseline risk times the relative reduction. Its reciprocal is the number needed to treat (NNT): how many people must be treated for one to benefit. ARR = baseline risk × RRR  ·  NNT = 1 ÷ ARRRRR = relative risk reduction · a "50% cut" in a 2% risk helps 1 in 100; in a 20% risk, 1 in 10 Interactive · Halving a risk, three ways Risk without treatment (%) Relative risk reduction (%) Each square stands for 10 people out of 1,000; filled squares show the events (rounded to whole squares, so very small risks show none; see the numbers below). Fictional treatment, round numbers. Benefits come with harms, which this simple picture leaves out. The same 50% cut helps 10 in 1,000 people when the baseline is 2% (NNT 100), and 100 in 1,000 when it is 20% (NNT 10). A real example of the other side: the Women's Health Initiative stopped its oestrogen-plus-progestin arm in July 2002 because harms exceeded benefits. A hazard ratio of 1.24 for breast cancer sounds large, but it meant about 8 extra cases per 10,000 women per year (in the long-run analysis), while pulmonary embolism rose with a ratio of 2.13, about 10 extra per 10,000, and colorectal cancer fell with a ratio of 0.56, about 7 fewer. The trial overturned the earlier observational impression of heart protection. Later analyses found that the picture varies with age and with how soon after menopause treatment starts; the decision today is individual and belongs in a conversation with a clinician. Screening and its harms Looking for disease in people with no symptoms sounds always good. The 1968 WHO principles by Wilson and Jungner set out the conditions: an important condition, a recognisable early stage, a treatment that works better early, an acceptable test. Screening also has costs. False positives mean worry, further tests and sometimes needless procedures. Overdiagnosis finds cancers that would never have caused harm in the person's lifetime, which cannot be known for any one patient. And the numbers behave unintuitively because of prevalence: Interactive · 1,000 people screened Prevalence of the condition (%) Sensitivity (%) Specificity (%) An illustrative test, not any real screening programme. PPV = positive predictive value. Overdiagnosis and harms of follow-up are not in this calculation. At 1% prevalence, 90% sensitivity and 95% specificity, a screened thousand has 9 true positives and about 50 false positives: only about 15% of positive results are real. Drop prevalence to 0.1% and it is under 2%. A positive result in a low-risk population is mostly a false alarm, which is the same Bayes logic as in 12/17, and why screening is offered to groups where the condition is common enough and not to everyone for everything. A second trap is lead-time bias. Screening finds disease earlier; survival is counted from diagnosis; so people appear to live longer even when death comes at the same time. Take ten invented patients who die 1, 2, 3, 3.5, 4, 4.5, 6, 8, 10 and 12 years after the point at which symptoms would have led to diagnosis. Without screening 4 of 10 (40%) survive five years from diagnosis. Detect everyone two years earlier and the same deaths give 8 of 10 (80%) without a single death delayed. The honest outcome for a screening test is deaths from the disease in a randomised comparison, not "five-year survival". Interactive · Survival that is only a shift in the clock Years earlier the disease is found Ten invented patients; the moment of death is fixed. Bars show time from diagnosis to death; the dashed line is five years. Systematic reviews and Cochrane: counting all the trials One trial is a data point. Many trials, found and combined by explicit rules, make a systematic review; if their results are pooled numerically, a meta-analysis. The idea, pushed by the British epidemiologist Archie Cochrane (his 1972 book Effectiveness and Efficiency argued that health services should rest on randomised evidence), led to the Cochrane Collaboration, formed in 1993 under Iain Chalmers. Cochrane reviews follow written protocols that say in advance which trials count. Critics of the movement note unpublished studies, slow updates and many reviews ending in "inconclusive". The simplest pooling gives each trial a weight of 1 ÷ its variance, so large, precise trials count most. A forest plot shows each trial's estimate and interval with a diamond for the pool. The pool is only as good as the trials found: if small studies with exciting results are published and small null ones are filed away (publication bias), the pool is rosy. Interactive · Pooling six invented trials All sixOnly the large twoOnly small "positive" ones Invented trials of a fictional treatment, effects as risk ratios (below 1 favours treatment). Fixed-effect inverse-variance pooling on the log scale. Real reviews also check for differences between trials and for bias. Pooled together the six trials give a risk ratio of about 0.85 (95% interval 0.76 to 0.96). Keep only the two large ones and it is 0.90, an interval of 0.79 to 1.03 that includes no effect. Keep only the three small trials that were "positive" (A, B and F) and the answer looks like a spectacular 0.62. Same drug, three answers, depending on which studies you let in. That is why reviewers search for unpublished work and register their protocols. Landmark trials and failures Medicine's history is a record of claims that felt true until someone counted. YearEventThe lesson 1747James Lind compares remedies for scurvy on a ship; the citrus pair recoversA fair comparison can be small and still clear; the lesson was ignored for decades 1847Ignaz Semmelweis, Vienna: chlorinated-lime handwashing; deaths in the First Clinic fell from about 18% in April to about 2% in JuneA before-and-after comparison with a dramatic effect, but no germ theory behind it, and not published by him until 1858–61. The tale that he was ignored out of pure stubbornness is simpler than the history 1948MRC streptomycin trial in tuberculosisThe first published RCT in medicine: allocation by chance, not by the doctor's hunch 1950Doll and Hill, smoking and lung cancerA case-control study, then a cohort, each confirming the other 1957–62Thalidomide, sold from 1957 in West Germany, withdrawn in Europe in November 1961 after birth defects in an estimated 10,000 to 20,000 childrenLed to the 1962 Kefauver-Harris amendments in the US: drugs must show efficacy and safety 1989–91CAST: drugs that suppressed abnormal heartbeats after heart attackThe drugs "worked" on the surrogate. In the trial 7.7% of the drug group died against 3.0% on placebo. A surrogate is not the outcome 2002Women's Health Initiative stopped earlyObservational studies had suggested a benefit that a randomised trial did not confirm The Tuskegee study (1932–72) is a failure of ethics rather than design; see Medical Ethics & Law (35/60). Evidence-based medicine itself has critics, who say that average trial results do not always apply to the individual in the room, that trial populations differ from real patients, and that industry funding can shape which questions get asked. Its defenders reply that the answer is to improve the evidence and judgment, not abandon either. What a graduate just sees Every percentage has a denominator; ask "out of how many?" before anything else. Relative changes sound large; absolute changes tell you what happens to a person. "Linked to" is a prompt to ask what else travels with the exposure. Only random assignment balances confounders you did not measure. An odds ratio is not a relative risk when the outcome is common. A positive result for a rare condition is more likely a false alarm than a case. Five-year survival can rise with no benefit; look for deaths in a randomised comparison. One trial is a data point; a review is only as good as the studies it finds. A good surrogate marker can still lead you wrong; check the outcome that matters. Where it shows up next Clinical Reasoning & Diagnosis (35/59): sensitivity, specificity and likelihood ratios are the maths of a test in a clinic. Public & Global Health (35/61): vaccination programmes, screening policy and disease burden rest on these methods. Immunology (35/57) and Pharmacology (35/56): the vaccine and peanut trials, and every drug approval. Medical Ethics & Law (35/60): consent, equipoise and when a trial may stop early. In the world: health technology agencies (NICE in England, IQWiG in Germany and others), clinical guidelines, drug regulators, health journalism. If you want to go deeper 📘 Textbook. Fletcher, Fletcher and Fletcher, Clinical Epidemiology: The Essentials (Wolters Kluwer). A short, clinical introduction used by many medical schools. 📘 Textbook. Trisha Greenhalgh, How to Read a Paper: The Basics of Evidence-Based Medicine and Healthcare (Wiley). The standard guide to appraising a published study. 📘 Gentler. Imogen Evans, Hazel Thornton, Iain Chalmers and Paul Glasziou, Testing Treatments (2nd edition 2011, foreword by Ben Goldacre), free to download at testingtreatments.org. Also Gerd Gigerenzer, Calculated Risks (2002). 🌐 Free course. Cochrane Interactive Learning: Module 1, introduction to conducting systematic reviews, is free with a login. The later modules need a subscription. The takeaway Medicine knows what works by counting, comparing fairly and then asking what was left out. Always ask: out of how many, compared with what, chosen how, and found by whom. Try this: next time a headline gives a percentage change in risk, find or estimate the baseline and work out the absolute change. If the article never says the baseline, that tells you something too. Sources & Further Reading Sources. Definitions and designs are standard epidemiology (Fletcher). MRC streptomycin (1948) as first published RCT, and Lind (1747): Wikipedia, Randomized controlled trial. Semmelweis figures: Wikipedia, Ignaz Semmelweis. Thalidomide: Wikipedia, Thalidomide. CAST (Echt et al., N Engl J Med 1989 preliminary report; 7.7% vs 3.0%, 56/730 vs 22/725). Women's Health Initiative: figures from Wikipedia's summary of the long-run analysis; stopped July 2002. Cochrane formed 1993: Wikipedia, Cochrane (organisation). Doll and Hill (1950, 1954) and the Wilson and Jungner WHO principles (1968) are standard history. All interactive numbers were computed in node. This is not medical advice. 📘 Greenhalgh, How to Read a Paper 📘 Fletcher, Fletcher and Fletcher, Clinical Epidemiology: The Essentials 🌐 Testing Treatments (free) 🌐 Cochrane Interactive Learning Step 11. Clinical Reasoning & Diagnosis: Probability With a Human in the Loop ----------------------------------------------------------------------------- Why: From symptoms to diagnosis: how clinicians think, and how they err. Doctors do not look up the answer; they move a probability with each fact, using a mind that is fast, usually right and predictably biased. How that works, and how it breaks. A doctor walks into a room, sits down and asks, "So what brings you in today?" Then she listens for a minute or two without interrupting, which is rarer than it should be. By the time she reaches for the stethoscope, she has already narrowed a hundred possible explanations to a handful. The tests she orders afterwards are not a fishing trip; each is meant to move a probability. That is diagnosis: not a flash of genius but a running estimate that rises and falls as evidence arrives, made by a human mind with a fast pattern-matcher and a slow checker, both of which err. All cases in this piece are invented, and every number is an illustration. This explains how medicine works and is not medical advice, and it is not a tool for diagnosing yourself or anyone else. For a health question, see a clinician; in an emergency, call your local emergency number. If you or someone you know is in a mental-health crisis, findahelpline.com lists services by country. The course in one box Who takes itMedical students, mainly in years 3 to 4 or as the bridge to clinical work, and increasingly nursing and physician-associate students. Often taught as "clinical reasoning", "clinical decision making" or inside the early clinical-skills course, and tied to the UK GMC's and the US AAMC's expected outcomes for graduates. What it is forTurning a story, an examination and a few tests into a sensible working diagnosis, and knowing how that process fails. Comes afterPathology (35/55), because you need to know what diseases look like, and Epidemiology & Evidence-Based Medicine (35/58), which gives sensitivity, specificity and the base-rate logic used here. Leads toClinical rotations and every specialty; closely related to Surgery, Anaesthesia & Emergency Care (35/62) and Clinical Psychology (35/27), where diagnosis follows other rules. What this piece gives youYou will not be able to diagnose anything. You will see why the history does most of the work, how a test result changes a probability, why a healthy person often has an "abnormal" result, how a clinician decides when a test is worth doing, and how smart people anchor on the wrong answer. Easier first reads: The Biases That Fool Everyone (17/02) and Bayes' Theorem (12/17). The story comes first: history, examination, tests A classic study by John Hampton and colleagues (British Medical Journal, 1975) followed 80 patients sent to a medical outpatient clinic. After the referral letter and the history, the clinician's diagnosis matched the final one in 66 of 80 cases, about 82%; the examination added little, and tests mostly confirmed. A later replication in Salt Lake City (Peterson and colleagues, 1992) found about 76%. These are small, old studies of a particular clinic, and they are the source of the saying that "listen to the patient, he is telling you the diagnosis", so treat the exact percentages as a rough indicator, not a law. The reason is simple. A story contains a lot of information: how fast it started, what makes it better or worse, what the person was doing, what else runs in the family. The examination tests specific ideas the history raised. Tests then confirm or refute them, and each should be chosen because its result would change what you do. A useful habit has a name: hypothetico-deductive reasoning. Form a short list of explanations early (often within the first minute), then ask what finding would separate them. Probability, not certainty: pre-test odds and likelihood ratios Before a test, the clinician has a pre-test probability: how likely this patient is to have the condition, given age, history, setting and the local base rate. A test then shifts it. The best summary of a test's power is its likelihood ratio (LR): how much more common the result is in people with the condition than in people without it. It comes from the two familiar numbers. LR+ = sensitivity ÷ (1 − specificity)  ·  LR− = (1 − sensitivity) ÷ specificitypost-test odds = pre-test odds × LR · odds = probability ÷ (1 − probability) The picture was put on a single chart by Fagan in 1975: a nomogram where a straight line from the pre-test probability through the LR gives the post-test probability. You can do the same with a slider. Interactive · Moving a probability with a test Pre-test probability (%) Test sensitivity (%) Test specificity (%) PositiveNegativePositive twice An invented test for an unnamed condition. "Positive twice" assumes two independent tests, which real tests rarely are. Illustration, not a calculator for any real test or person. Take a pre-test probability of 10% and a test with 90% sensitivity and 95% specificity. The LR+ is 18 and the LR− is 0.105. A positive result moves 10% to about 67%; a negative one moves it to about 1.2%; two independent positives would take it to 97%. Notice what a positive did not do: it did not make the answer certain. And with a pre-test probability of 1%, the same positive only reaches about 15%, which is the screening lesson of Epidemiology (35/58) in another form. Many tests, one patient: why "normal" is a range A reference range is usually set to include the middle 95% of healthy people. That means 5% of perfectly healthy people fall outside it on any one test, by definition. Order several tests and the chance that at least one is flagged climbs. P(at least one flag) = 1 − cnc = share of healthy people inside the range for one test · n = number of independent tests Interactive · A panel of tests on a healthy person Number of independent tests Share of healthy people inside the range (%) Independent tests in a healthy person. Real results are partly correlated, which lowers the chance somewhat, but not to zero. Illustrative. With 20 independent tests and 95% ranges, a healthy person has about a 64% chance of at least one flag; with 40, about 87%. Widen each range to cover 99% of healthy people and 20 tests flag at least one in only 18% of cases. This is why clinicians do not order "everything" for someone who feels well: the abnormal results are more often noise than disease, and each one invites a follow-up with its own harms. A test is ordered when the result would change what happens next. The differential and the treatment threshold The differential diagnosis is the ranked list of explanations still in play. Two proverbs frame it. "When you hear hoofbeats, think horses, not zebras" (attributed to a Maryland professor, Theodore Woodward, in the 1940s) says to start with what is common. Hickam's dictum answers that a patient can have as many diseases as they please. A third rule overrides both: always ask what you must not miss. A rare diagnosis that can kill within hours and is easy to treat earns a place on the list even at a low probability. That last rule is a number. In 1980 Stephen Pauker and Jerome Kassirer described the threshold approach: you should treat when the probability of disease passes the point where expected benefit to those who have it equals expected harm to those who do not. treatment threshold = harm ÷ (harm + benefit)harm = cost of treating someone without the disease · benefit = gain from treating someone with it, in the same units If the treatment is gentle and the disease dangerous, the threshold is low and you treat on a hunch. If the treatment is risky and the disease mild, the threshold is high and you want more certainty. Between a lower and an upper probability a test is worth doing, because a positive could push you over the threshold or a negative could pull you under it. Interactive · When is a test worth doing? Harm of treating someone without the disease Benefit of treating someone with it Your current probability (%) Harm and benefit are in the same invented units; the test is an invented one with LR+ 10 and LR− 0.1. A teaching picture of the Pauker and Kassirer idea, not a decision tool. With harm 5 and benefit 45, the treatment threshold is 10%. A test that moves odds ten-fold up or down is worth doing when the probability is between about 1.1% and 52.6%. Below it, even a positive result would stay under the threshold. Above it, even a negative result would stay over it. Change the harm to 20 and the benefit to 20: the threshold becomes 50% and the testing zone runs from about 9% to 91%. The bigger the harm of treatment relative to the benefit, the more evidence you need first. Two ways of thinking, and how each fails Experienced clinicians often diagnose in seconds: the "look" of the patient or the pattern of the story matches an illness script built from hundreds of cases. This is fast, usually right and hard to explain, which psychologists call System 1 (see The Biases That Fool Everyone, 17/02, for the general idea). The other mode is slow and analytic: list the possibilities, test them against the findings, think about what would make them wrong. Researchers such as Pat Croskerry have argued that most diagnostic errors involve a failure of the fast mode that the slow mode did not catch; others, notably Geoff Norman and colleagues, point out that experts also err when they are slow and that knowledge gaps matter. The field has not settled which "fix" works best. The named failures are well known. Anchoring: sticking with the first idea. Premature closure: stopping once a plausible answer appears. Confirmation bias: giving more weight to findings that fit. Availability: overestimating what you saw last week. Base-rate neglect: forgetting how common the condition is. Debiasing, such as the habit of asking "what else could this be?" or "what is the worst thing it could be?", has some support but mixed evidence in trials. Try an invented case with fictional diseases. Interactive · Anchoring on a first impression Invented patient "Ms Ardent". Three fictional possibilities: common, mild Brannock fever (prior 90%), serious but rare Tolliver's disease (5%), and something else (5%). Reveal findings one at a time. Weight given to findings that go against your first idea (%) Reveal next findingStart again All diseases, findings and numbers are invented to show the process. The "anchored" bar discounts findings that favour the other disease by the weight you set. Not a model of any real clinician. Weighed fully, the findings (fever, mild cough, swollen joints, then a strongly positive Test Z) take Tolliver's disease from 5% to 74%. A clinician who gives only 30% weight to findings against Brannock fever ends the same case at about 9% Tolliver, 90% Brannock: the story is still "a common mild illness". The error did not need a wrong fact; it needed only a discount on the awkward ones. A simple defence is to say out loud, "what finding would make me change my mind?" and then check whether it has appeared. How often is diagnosis wrong? Because diagnosis is a process, error is hard to count: it needs a gold standard such as an autopsy or follow-up. The US National Academy of Medicine's 2015 report Improving Diagnosis in Health Care defined diagnostic error as the failure to establish an accurate and timely explanation of the patient's health problem, or to communicate it to the patient, and concluded that most people will experience one at least once. A review by Berner and Graber (2008) put erroneous physician diagnoses at about 10 to 15%. A 2023 modelling study led by David Newman-Toker at Johns Hopkins (published in BMJ Quality & Safety) estimated that about 795,000 Americans a year die or are permanently disabled by diagnostic error: 371,000 deaths and 424,000 disabilities, with vascular events, infections and cancers accounting for three-quarters. That is an estimate with wide uncertainty from one country, built on models, and earlier estimates ranged from 40,000 to 4 million, so treat it as an order of magnitude and not a count. The causes are mixed: faulty thinking, but also missing information, busy rooms, hand-offs, tests not followed up, and patients who were not heard. Error is as much a matter of systems as of individuals, which is why patient safety work looks at how a hospital runs, not only at who was tired. AI in diagnosis: strong at pictures, shaky at context Computers can now match specialists on some narrow tasks. In 2018 the US FDA authorised the first AI system to diagnose a specific eye disease, diabetic retinopathy, from retinal photographs. Systems read mammograms, skin photographs and scans, though a 2020 claim that an algorithm could beat expert radiologists drew replies from researchers who said the paper lacked the detail needed to reproduce it. For general reasoning, large language models can score well on case vignettes. In a 2024 randomised trial in JAMA Network Open (Goh and colleagues), physicians given access to an LLM scored 76% on diagnostic reasoning against 74% for those using usual resources, a difference that was not statistically significant, while the model alone scored higher than both groups. A good tool on its own is not the same as a good tool in a doctor's hands. The limits are specific. A model trained at one hospital can fail at another with different scanners, patients or prevalence, which is called dataset shift. Few studies reporting AI accuracy have been tested outside the place they were built. A widely used commercial sepsis-warning score, when tested at one US university hospital system (Wong and colleagues, JAMA Internal Medicine, 2021), had an area under the curve of 0.63, missed 67% of patients who developed sepsis and still raised an alert on 18% of hospitalisations. And people tend to over-trust a confident machine (automation bias). The base-rate arithmetic from earlier still applies: Interactive · Moving a good AI to a new place Prevalence where it is used (%) Specificity lost in the new place (points) An invented AI with 92% sensitivity and 90% specificity where it was built (prevalence 20%). Numbers per 1,000 people flagged or not. Illustration, not any real product. Where it was built, at 20% prevalence, 70% of its alerts are real (184 true, 80 false per 1,000). Used at 1% prevalence with the same accuracy, only 8.5% of alerts are real (about 9 true and 99 false per 1,000). Lose 10 points of specificity in the new setting and it falls to 4.4%. The software did not get worse; the world around it changed. Most clinical AI is therefore best judged by prospective tests in the setting where it will be used, and by whether patients do better. What a graduate just sees The history does most of the work; tests and examination test ideas the story raised. Every diagnosis is a probability that moves with evidence; certainty is rare. A test result is only as meaningful as the pre-test probability it lands on. "Normal" is a range, and healthy people fall outside it more often as the panel grows. Keep a short list in mind, include what you must not miss, and ask what would separate the options. The treatment threshold depends on harm and benefit, not only on probability. Fast pattern recognition is mostly right, and its errors come from not stopping to check. Ask: "What else could this be? What would change my mind?" Diagnostic error is common, and as much a system problem as a personal one. A model that works in one hospital can fail in another, because prevalence and data change. Where it shows up next Pathology (35/55): the patterns of disease that illness scripts store. Epidemiology & Evidence-Based Medicine (35/58): sensitivity, specificity, screening and the base-rate trap. Surgery, Anaesthesia & Emergency Care (35/62): decisions under time pressure, where thresholds drop. Medical Ethics & Law (35/60): disclosing uncertainty and error. In the world: triage, radiology and pathology reporting, clinical decision support, second-opinion services and patient-safety programmes. If you want to go deeper 📘 Textbook. Jerome Groopman, How Doctors Think (2007) is a good start; for a course text, Scott Stern, Adam Cifu and Diane Altkorn, Symptom to Diagnosis: An Evidence-Based Guide (McGraw Hill). It works through presenting complaints with likelihood ratios. 📘 Textbook. Harold Sox, Michael Higgins and Douglas Owens, Medical Decision Making (Wiley-Blackwell). The test-and-threshold maths, in more depth. 📘 Gentler. Daniel Kahneman, Thinking, Fast and Slow (2011), for the two-systems background, and the National Academies report Improving Diagnosis in Health Care (2015, free to read online at nap.edu). 🌐 Free course. Khan Academy's health and medicine courses are free; for the maths of tests, the free Testing Treatments book (also in 35/58) is a gentle companion. The takeaway Diagnosis is a probability that moves with each piece of evidence, steered by a mind that is fast, usually right and predictably biased. A test is worth doing when its result could change what happens next. The best defences are a short list, a worry about what you must not miss, and the question "what would change my mind?" Try this: next time you make a confident guess about anything (why the car will not start, why a friend is late) write down your first idea and two others, and the one finding that would make you drop the first. Then see whether you look for it. Sources & Further Reading Sources. Likelihood ratios, Bayes in diagnosis and the threshold idea (Fagan 1975; Pauker and Kassirer 1980) are standard clinical-decision-making teaching. History-taking: Hampton et al., BMJ 1975 (66 of 80); Peterson et al., West J Med 1992 (about 76%), via secondary summaries. Diagnostic-error definition: National Academy of Medicine (2015); Berner and Graber (2008), via Wikipedia, Diagnostic error. Newman-Toker et al., BMJ Qual Saf 2024;33:109–120 (online July 2023). Goh et al., JAMA Netw Open 2024;7(10):e2440969. Wong et al., JAMA Intern Med 2021 (AUC 0.63; 67% of sepsis missed; alerts on 18%), via PubMed. FDA 2018 retinopathy authorisation: Wikipedia, Artificial intelligence in healthcare. All cases are invented; all interactive numbers were computed in node. This is not medical advice. 📘 Groopman, How Doctors Think 📘 Stern, Cifu, Altkorn, Symptom to Diagnosis 📘 Kahneman, Thinking, Fast and Slow 🌐 Testing Treatments (free) Step 12. Medical Ethics & Law: Whose Body, Whose Decision --------------------------------------------------------- Why: Consent, confidentiality, end of life and the law. Every clinic runs on a few quiet promises: that you decide what happens to you, that what you say stays private, that the research behind your treatment was fairly done, and that scarce care is shared by rules. Medical ethics is the study of those promises and of what to do when they collide. A man in his seventies is told his scan shows a cancer. His daughter takes the doctor aside: "Please don't tell him. It will destroy him." Whichever way the doctor goes, she is answering a question that no blood test can answer: whose information is this, and who decides? Medical ethics is what a graduate learns to see in a moment like that. It is a small set of ideas for looking at a hard case, plus a history of what went wrong when they were ignored. This piece explains how medical ethics and law work. It is not medical or legal advice; laws differ by country and change, and for a health question you should see a clinician. In an emergency, call your local emergency number. The course in one box Who takes itMedical, nursing and allied-health students, usually woven through the years of training and examined late (year 3 onward). Philosophy, law and public health students meet it as bioethics. What it is forGiving clinicians and patients a shared language for decisions where the medically best option, the person's wishes, the law and fairness pull in different directions. Comes afterPolitical Theory (35/12), for the ideas of liberty, justice and legitimacy it borrows. Leads toPublic & Global Health (35/61), where the same questions apply to populations, and Surgery, Anaesthesia & Emergency Care (35/62), where consent is asked in a hurry. It also sits behind Epidemiology & Evidence-Based Medicine (35/58) and Clinical Psychology (35/27). What this piece gives youYou will not be able to say what the law allows in a particular case; it differs by place and changes. You will see the four principles and their critics, what capacity really tests, why confidentiality has edges, how research scandals wrote the rules, how end-of-life law differs across countries, and why rationing is unavoidable. Easier first reads: Justice: Plato to Rawls, Nozick and Sen (05/13), Moral Philosophy, Seriously (05/15) and Rights, Constitutions and the Rule of Law (18/04). The four principles, and the people who push back In 1979 two American philosophers, Tom Beauchamp and James Childress, published Principles of Biomedical Ethics. It proposed four principles as a common vocabulary for clinicians who disagree about theory: respect for autonomy (people decide about their own lives), beneficence (act for the patient's good), non-maleficence (avoid causing harm) and justice (distribute benefits and burdens fairly). The book is in its eighth edition (2019) and is the most-used framework in medical teaching. The authors are clear that the principles do not rank themselves. Each is a strong reason that can be outweighed; you must specify what a principle means in a context and then balance it against the others. That is where the argument starts. Critics say this gives no way to settle conflicts: Danner Clouser and Bernard Gert (1990) called the principles a checklist of unconnected slogans, a "pick and mix" of different moral theories. Virtue ethicists say it ignores the character of the person acting; casuists say reasoning from paradigm cases beats principles. Defenders such as Raanan Gillon reply that the four principles are common ground between people of many moral views, which is exactly why they are useful at a bedside. Interactive · Same facts, different weights Refuses bloodDon't tell himOne bed left Weight on autonomy Weight on beneficence Weight on non-maleficence Weight on justice Illustration only. Each option is scored 0 to 3 on how well it honours each principle, by the author, to show that the answer moves with the weights. Beauchamp and Childress themselves reject a formula like this; real cases are settled by argument about facts, not arithmetic. With equal weights of 5, respecting a competent adult's refusal of blood scores 40 against 30 for transfusing anyway. Put the weight on welfare (beneficence and non-maleficence at 10, autonomy and justice at 2) and transfusion edges ahead, 44 to 40. The sliders do not give the right answer; they show where the disagreement lives, usually in how much a person's choice should count, not in the facts. Consent and capacity: whose decision is it? The legal core of autonomy is informed consent. In 1914 Judge Benjamin Cardozo wrote in Schloendorff v. Society of New York Hospital that every adult "of sound mind has a right to determine what shall be done with his own body"; an operation without consent is a wrong even if it is skilful. Since then the question has been what must the patient be told. The older English answer was what a responsible body of doctors would say (Bolam, 1957); Montgomery v Lanarkshire Health Board (UK Supreme Court, 2015) made it the risks this patient would want to know. A consent only counts if the person can make the decision. Capacity is not a label on a person; it is decision-specific and time-specific. In England and Wales the Mental Capacity Act 2005 (in force 2007) starts from five rules: assume capacity; give all practicable help first; an unwise decision is not proof of incapacity; if someone lacks capacity, act in their best interests; choose the least restrictive option. Its test asks whether the person can understand the relevant information, retain it long enough, use or weigh it, and communicate a choice. Other systems word it differently but ask something very similar. capacity = understand + retain + use or weigh + communicatefor one decision, at one time; failing any part means the person lacks capacity for that decision Interactive · Does she have capacity? Invented vignettes for illustration; capacity assessments belong to trained clinicians and, in the last resort, courts. If an eating disorder or any mental-health struggle touches you, findahelpline.com lists services in your country. The third case is where the field itself is unsettled. Clinicians and courts disagree about when a belief that comes from an illness stops a person from weighing the facts, even though they can repeat them. That is a live debate, not an error in the test. Confidentiality and where it stops "Whatever I see or hear in the lives of my patients ... I will keep to myself": the line is from the Hippocratic tradition, and some version of the promise appears in oaths from ancient Greece to the Indian Charaka Samhita to modern declarations. The reason is practical: people who fear exposure hide symptoms. But no legal system treats it as absolute. Disclosure is usually allowed or required in four kinds of situation: the patient agrees; the law requires a report (notifiable infectious diseases in most countries, certain injuries, suspected child abuse); a court orders it; or there is a serious risk to someone else that cannot be avoided otherwise. The best-known case for the last is Tarasoff v. Regents of the University of California. In 1976 the California Supreme Court held (17 Cal. 3d 425) that a therapist who reasonably believes a patient is a serious danger to an identifiable person has a duty to take reasonable steps to protect that person. US states have since followed it in different ways, from mandatory to merely permitted; elsewhere, professional bodies such as the UK's General Medical Council allow disclosure in the public interest. Interactive · Who may be told? General patterns only; the exact rule depends on the country, the profession and the facts. Not legal advice. Research ethics: the rules were written after the scandals At the Doctors' Trial in Nuremberg, American judges tried Nazi physicians for lethal and torturous experiments on camp prisoners. The 1947 judgment set out ten principles, the Nuremberg Code; the first begins, "The voluntary consent of the human subject is absolutely essential." In 1964 the World Medical Association adopted the Declaration of Helsinki, which put research ethics in doctors' own hands; it has been revised many times, most recently in October 2024, which added emphasis on global inequities and scientific integrity. Rules written by doctors did not stop the next scandal. From 1932 to 1972 the US Public Health Service, with Tuskegee Institute, followed about 600 poor Black men in Macon County, Alabama: 399 with syphilis and 201 without. By 1947 penicillin was the standard treatment, and the men were still not given it. Whistleblower Peter Buxtun went to the press; the story broke on 25 July 1972 and the study ended that November. A 1974 class action brought a $10 million settlement, and President Clinton apologised on 16 May 1997. Congress passed the National Research Act in 1974, which led to ethics review boards, and in 1979 the Belmont Report named three principles for research: respect for persons, beneficence, and justice (who bears the burden and who gets the benefit). A different kind of case is Henrietta Lacks. In 1951, at Johns Hopkins Hospital in Baltimore, doctors took samples of her cervical cancer without asking her; she died on 4 October 1951, aged 31. Her cells became the first human cell line that kept dividing, HeLa, used worldwide for decades. Her family learned of it in 1975; the HeLa genome was published without their consent in 2013, after which the NIH and the family agreed on controlled access to the data. Rebecca Skloot's 2010 book told the story, and in July 2023 the family settled a claim against a biotech supplier on undisclosed terms. At the time, permission was neither required nor customarily sought; in Moore v. Regents (1990), the California Supreme Court later held that discarded tissue is not a person's property. Law and ethics did not move together. Interactive · Scrub through the years Year Red dots are wrongs and revelations; blue dots are rules and rulings. The end of life: refusing, withdrawing, dying with help Three things are often run together. Refusing or stopping treatment is allowed for a person with capacity in most legal systems, even when death follows; it is why advance directives exist. Euthanasia means a clinician gives a drug that ends the person's life; assisted suicide means the clinician prepares or prescribes and the person takes it. Pain relief that might hasten death as a side effect is judged under the older "double effect" idea. The laws on euthanasia and assisted suicide differ widely and are changing quickly. The Netherlands and Belgium (2002) were the first to legalise euthanasia by statute. Switzerland never legalised it but has not punished assisting a suicide unless the motive is selfish since 1942. Colombia's Constitutional Court decriminalised it in 1997, and Canada, Spain, New Zealand and every Australian state have passed laws since 2016. Arguments for the laws stress autonomy and relief from suffering. Arguments against stress risks to disabled and vulnerable people, the quality of palliative care, and the role of doctors as healers. Table · Some laws, with how they came about AllStatuteCourt rulingNot yet in force PlaceWhat it allowsSince SwitzerlandAssisted suicide, self-administered; not a crime unless the motive is selfishPenal Code art. 115, in force 1942 NetherlandsEuthanasia and assisted suicide, with review committees1 April 2002 BelgiumEuthanasia2002 LuxembourgEuthanasia2009 ColombiaEuthanasia; the court decriminalised it, rules followed, widened in 2021Court 1997 CanadaMedical assistance in dying2016 (Royal Assent 17 June) SpainEuthanasia and assisted suicide25 June 2021 AustraliaVoluntary assisted dying, state by state: Victoria first, then WA, Tasmania, Queensland, South Australia, NSW2019 to 2023 New ZealandAssisted dying7 November 2021 Oregon (US)Assisted suicide only; euthanasia illegal; several other US states have similar laws1997 GermanyFederal Constitutional Court struck down the ban on organised assisted suicide; active euthanasia stays illegalCourt 2020 EcuadorConstitutional Court decriminalised euthanasia under conditionsCourt 5 Feb 2024 IndiaSupreme Court: withdrawing life support allowed; active euthanasia illegalCourt 2018 UruguayEuthanasia for adults with terminal illness; law passed Oct 2025, decree reported April 20262025 to 2026, check PortugalLaw 22/2023 passed but not in force as of May 2026Pending FranceAssembly gave final approval on 15 July 2026; reported promulgated in August after constitutional review; check implementationPending England and WalesAssisting suicide remains a crime; a bill was reintroduced in 2026, second reading scheduled 11 September; outcome not verifiedPending Checked on 1 November 2026 against Wikipedia's "Legality of euthanasia" and news reports; the 2026 entries may have moved. Always check current law. Scarce resources: who gets the last bed No health system can give everyone everything. Rationing is not a question of whether but how: by price, by waiting, by clinical rules, by lottery, or by a body that decides what the public purse will fund. Norman Daniels's "accountability for reasonableness" says a fair process publishes its reasons, lets people appeal, and revises. In the first waves of COVID-19, hospitals in many countries wrote ventilator triage protocols, and the criticism that a rule favouring likely survivors quietly discriminates against disabled people showed how hard it is to be efficient and fair. Health economists use the quality-adjusted life year (QALY): a year in full health counts 1, a year in a worse state counts less, death 0. Dividing the extra cost of a treatment by the extra QALYs it gives yields a cost per QALY, compared with a threshold a health system is willing to pay. England's NICE has long used a range around £20,000 to £30,000 per QALY (check the current figure). Critics say QALYs undervalue life with disability; defenders say that without some measure, the loudest case wins. cost per QALY = extra cost ÷ extra QALYsQALYs = years × quality of life (0 to 1) Interactive · A fictional treatment against a threshold Years of life with treatment Quality of life with treatment (0 to 1) Extra cost (thousand units) Threshold (thousand units per QALY) Fictional drug. Without treatment, 2 years at quality 0.5 (1.0 QALY). Money units are arbitrary. Not a funding decision. With the sliders at their starting values the treatment gives 4 × 0.6 = 2.4 QALYs against 1.0, a gain of 1.4, for 60 thousand units: about 43 thousand per QALY, above a threshold of 30. Raise the quality to 0.8 and the gain is 2.2, the cost per QALY drops to about 27, and the answer flips. The number is only as good as the quality weights, which are value judgments. Other ways to think about it The four principles grew from American liberal thought, and they travel unevenly. Scholars in other traditions have asked what a bedside looks like when the unit is not the lone chooser. In ubuntu thinking, widely discussed in southern Africa and summed up in the Nguni saying that a person is a person through other people, decisions are shared with family and community, and a diagnosis may belong to the family before the patient. Confucian bioethics, discussed by scholars in China, Korea, Japan and Taiwan, often places the family at the centre and sees a decision made together as a form of respect, not a loss of autonomy. Care ethics, from Carol Gilligan and Nel Noddings, starts from dependence. Islamic medical ethics draws on the aims of the law (protecting life, faith, intellect, lineage and property) and on fatwas. Latin American bioethicists, including the Brazilian school around Volnei Garrafa, argue that bioethics must start from inequality and poverty, not only consent forms. UNESCO's 2005 Universal Declaration on Bioethics and Human Rights tried to find common ground across these. These are broad traditions, not national stereotypes: within every country, patients and doctors differ, and many families abroad want what the Anglo-American texts call autonomy. The lesson: ask the patient how they want to decide, and with whom. What a graduate just sees Any hard case becomes four questions: whose choice, what helps, what harms, what is fair. The answer is in how they are weighed. Capacity belongs to one decision at one time; an unwise choice is not an incapable one. Confidentiality is a strong default with known edges: consent, the law, a court, a serious danger to others. Almost every research rule is a scar from an earlier scandal. Withdrawing, withholding, helping and doing are not the same act in law, and the line moves by country and decade. Rationing always happens; the question is whether it is done openly and with reasons. Autonomy as drawn in textbooks is one culture's emphasis; ask patients how they want to decide. Where it shows up next Epidemiology & Evidence-Based Medicine (35/58): trials need consent and review boards; the Belmont principles decide who may be enrolled. Public & Global Health (35/61): quarantine, vaccination mandates and rationing apply the same principles to populations. Surgery, Anaesthesia & Emergency Care (35/62): consent in a hurry, and in emergencies with no one to ask. Genetics (35/46) and Clinical Psychology (35/27): shared genes and compulsory treatment raise the hardest confidentiality and capacity questions. In the world: hospital ethics committees, the courts, research ethics boards, health-data law, and every conversation about a living will. If you want to go deeper 📘 Textbook. Tom Beauchamp and James Childress, Principles of Biomedical Ethics (8th ed., Oxford University Press, 2019). The framework itself and the objections to it. 📘 Gentler. Tony Hope, Medical Ethics: A Very Short Introduction (Oxford University Press). A short, readable tour of consent, capacity and rationing. Also Rebecca Skloot, The Immortal Life of Henrietta Lacks (2010). 📘 On dying. Atul Gawande, Being Mortal (2014), a surgeon on what matters at the end of life. 🌐 Free reading. The Declaration of Helsinki and UNESCO's Universal Declaration on Bioethics and Human Rights; free online health-ethics and research-ethics courses are also hosted at OpenWHO. The takeaway Medical ethics is four old questions (whose choice, what helps, what harms, what is fair) asked again with the facts of each case, and a body of law that records what societies decided after learning the hard way. The answers move with place and time, so the useful habit is to ask the questions out loud. Try this: this week, think of the last time you or someone you know had a medical decision to make. Who was told? Who decided? Was it a choice or an instruction? Notice which of the four questions was missing. (This is reflection, not advice.) Sources & Further Reading Sources. Beauchamp and Childress (1979, 8th ed. 2019); Clouser and Gert (1990); Gillon. Mental Capacity Act 2005 (England and Wales; royal assent 7 April 2005, in force 2007). Schloendorff (1914), Bolam (1957), Montgomery (2015), Moore v. Regents (1990) and Tarasoff (1976, 17 Cal. 3d 425) as standard legal references; Tarasoff variations from an encyclopaedia summary. Tuskegee: US Public Health Service 1932 to 1972, 600 men, 399 with syphilis; AP story 25 July 1972; settlement of $10 million in 1974; Clinton apology 16 May 1997; Belmont Report 1979 (encyclopaedia summaries of primary sources). Lacks: 1951, 1975, 2013, 2023 (encyclopaedia summary). Declaration of Helsinki: 1964, eighth revision October 2024. End-of-life laws: Wikipedia "Legality of euthanasia" (read 1 November 2026) and news reports for France and England. NICE range as widely reported (check current). The interactives' numbers were computed in node. Not medical or legal advice. 🌐 WMA Declaration of Helsinki 🌐 UNESCO bioethics 📘 Beauchamp and Childress, Principles of Biomedical Ethics 📘 Gawande, Being Mortal Step 13. Clinical Psychology: Where Normal Ends and Care Begins --------------------------------------------------------------- Why: Mental illness and its treatment. What makes something a disorder, how the two big manuals disagree, why causes are never just one thing, which treatments have good evidence, and why most of the world's people who need help never get it. This piece describes how clinical psychology works. It is not medical advice, and nothing here can tell you whether you or someone you know has a condition. If you are struggling, a doctor or a local mental health service is the right place to start; findahelpline.com lists free helplines by country. Everyone has had a week when they could not sleep, felt flat, or worried so much that they could not think. Almost nobody thinks that week is an illness. Yet there are people for whom that week lasts a year, and who cannot work, eat or be with their family. Somewhere between these lies a line, and it is not a line that a blood test can draw. That is the central puzzle of clinical psychology: how to tell ordinary suffering from a disorder, how to describe what is wrong without pretending to know more about the cause than we do, and how to help. A graduate of this course keeps three habits: treat diagnoses as useful descriptions with fuzzy edges, treat causes as many things at once, and ask of every treatment how well it was tested. The course in one box Who takes itPsychology majors, usually in year 3; it is the gateway to clinical training (a doctorate or a master's degree plus supervised practice in most countries, the route and the title differ by country). Nursing, social work and medical students take versions of it. What it is forDescribing, understanding and treating mental disorders, and knowing how strong the evidence is. Comes afterBiological Psychology (35/20) and Personality & Individual Differences (35/26). Leads toClinical and counselling training; it also feeds Organizational Behavior (34/53) and, with Developmental Psychology (35/24), the study of how problems begin. What this piece gives youYou will not be able to diagnose anyone, and this piece does not offer a self-test. You will know how a disorder is defined, what the main conditions are, what works and how well, and what the global picture looks like. Easier first reads: How to Actually Calm Down (01/03), Stress vs Burnout (01/05) and Aging Well (02/10). What makes something a disorder The definition in the American Psychiatric Association's manual runs roughly like this: a mental disorder is a clinically significant disturbance in thinking, emotion regulation or behaviour that reflects a problem in psychological, biological or developmental processes, and that usually comes with distress or disability. It adds exclusions: an expected response to a loss is not a disorder, and neither is behaviour that is merely socially deviant, or in conflict with society, unless it reflects a dysfunction in the person. Philosophers have argued about this for decades. Jerome Wakefield's "harmful dysfunction" idea says a disorder needs both a failure of some mental mechanism to do what it evolved to do and harm to the person. Critics ask who decides what a mechanism is for. In practice, clinicians use three working questions: how much does it distress the person, how much does it get in the way of life, and how far is it outside what the person's culture expects? The history shows how much the line moves. Homosexuality was listed as a mental disorder in early editions of the American manual; the American Psychiatric Association's board voted to remove it in 1973, and the World Health Organization removed it from the international list in 1990. In 1973 David Rosenhan published "On Being Sane in Insane Places", reporting that eight people who feigned hearing a voice were admitted to psychiatric hospitals; the paper was influential, and Susannah Cahalan's 2019 book The Great Pretender has since cast serious doubt on how the study was run and reported. Most symptoms form a continuum. Plot how much sadness or worry people have and you get a bell curve with no natural gap; a diagnosis puts a line through it. Move the line and the number of cases moves with it. Interactive · Where do we draw the line? Cut-off, in standard deviations above average An invented, normally distributed "symptom load"; real conditions are not exactly bell-shaped and real diagnosis also asks about impairment and duration. This is a picture of an idea, not a screening tool. Without JavaScript: a cut-off 1.5 standard deviations above average labels about 6.7% of people; at 2 about 2.3%; at 1 about 15.9%. Whoever stands just below the line and whoever stands just above it are nearly alike. That is why a diagnosis is useful for getting help and for research, and why clinicians stress that it describes a pattern, not a separate kind of person. DSM-5-TR and ICD-11: two manuals, one field Two systems define what is counted. The DSM is published by the American Psychiatric Association; its fifth edition appeared in 2013 and a text revision, DSM-5-TR, in March 2022. The ICD is the World Health Organization's classification of all diseases; its eleventh revision, ICD-11, was endorsed by the World Health Assembly in May 2019 and came into effect on 1 January 2022, with a chapter on mental, behavioural and neurodevelopmental disorders. ICD is the official statistical standard for WHO member states, and many countries are still moving to ICD-11; DSM is used mainly in the United States and in much research around the world. They are mostly alike, but they differ at the edges, and the edges are revealing. Interactive · Same condition, two manuals Simplified summaries. Check the manuals for exact criteria. DSM-5-TR (APA, 2022)ICD-11 (WHO, in effect 2022) Who uses itMainly the US; widely in researchHealth systems and statistics worldwide StructureCategories with symptom checklistsClinical descriptions, with more dimensions CostSold as a bookFree online Do the manuals agree on who has what? Not as well as people assume. In the field trials for DSM-5, reported by Regier and colleagues in 2013, the agreement between clinicians interviewing the same patients on separate occasions was low for some common conditions, including major depression and generalized anxiety disorder, a result the APA itself acknowledged. Diagnoses are tools with real uncertainty, and critics have proposed more dimensional systems, such as the HiTOP model (Hierarchical Taxonomy of Psychopathology), which describes problems as positions on a few broad scales. The main conditions, in outline This is a map, not a guide to identifying anything. The two most common groups worldwide are anxiety and depressive disorders: the World Health Organization estimated in 2023 that about 1.2 billion people, roughly one in seven, were living with a mental disorder, including about 470 million with anxiety and 322 million with depression. Most conditions begin before the mid-twenties. GroupThe core of itMain treatments with evidence DepressivePersistent low mood or loss of interest, with sleep, energy and thinking changesPsychotherapy, medication, or both Anxiety, OCD, PTSDFear and worry out of proportion; obsessions and compulsions; reactions after traumaCBT with exposure; some medicines Bipolar and psychoticEpisodes of mania; losing contact with shared reality (psychosis)Medication is central, with psychosocial care NeurodevelopmentalAutism, ADHD, intellectual disability: differences present from early lifeSupport and adjustment, with medication for ADHD Eating and substance useDisordered eating; harmful use that is hard to stopSpecialist psychological, medical and community care PersonalityLong-standing patterns that cause trouble with self and othersSpecialist psychotherapy (for example, DBT) A diagnostic test is a screen of a different kind of question, and one piece of arithmetic helps whoever reads about any test. The chance that a positive result is a true case depends heavily on how common the condition is. This is Bayes' rule at work, the subject of Bayesian Thinking for Everyday Life (17/07). Interactive · What a positive result means How common the condition is Sensitivity: true cases the test catches Specificity: non-cases the test clears A made-up test for a made-up condition, applied to 1,000 people. It is about reasoning, not about any real instrument. Without JavaScript: with a 2% condition and a test that is 90% sensitive and 90% specific, about 16 of every 100 positive results are true cases; at 20% prevalence, about 69. This is why a short questionnaire is a first step, never a diagnosis: a positive result in a general population mostly leads to a conversation with a trained person, not a conclusion. Causes: the biopsychosocial picture "What causes depression?" has no single answer. The biopsychosocial model, introduced by George Engel in 1977, says that biology (genes, brain, illness), psychology (thinking habits, history) and the social world (poverty, relationships, culture) all matter, and they interact. The diathesis–stress model is the simplest version: a person has some vulnerability (the diathesis), and a disorder appears when stress pushes the total over a threshold. Two people with the same stress can have different outcomes. Twin studies suggest that genes matter, though not as destiny: heritability is high for schizophrenia (roughly 80%) and moderate for major depression (roughly 40%), and both leave plenty to the environment. Childhood adversity, loss, discrimination and poverty raise the risk of many conditions. And no single gene or chemical explains any common disorder. The "chemical imbalance" story deserves care. The serotonin theory of depression was popular for decades, and a 2022 umbrella review by Joanna Moncrieff and colleagues concluded that the research did not show a consistent link between serotonin and depression. Critics said the review was one-sided in its method and that it did not show that antidepressants do not work, and the researchers disagreed with them. What most clinicians now say is that medicines can help, that we do not understand exactly how, and that "depression is low serotonin" is a simplification. Interactive · Vulnerability plus stress Stress load Average vulnerability in the group A toy: vulnerability varies between people (standard deviation 15), and a disorder appears when vulnerability plus stress passes 100. The units are invented; the shape is the point. Without JavaScript: with average vulnerability 50 and stress 40, about 25% cross the threshold; at stress 60, about 75%. Look at the curve: stress is not a switch. It moves a share of the group over the line, and the people just below the line at one level of stress are the people over it at the next. That is both why prevention matters and why no person's illness is simply their own fault. Therapies and their evidence Psychotherapy has three broad families. Cognitive-behavioural therapy (CBT), developed from the 1960s by Aaron Beck and others, works on the link between thoughts, feelings and behaviour, and on facing feared things in steps; it has the largest research base. Psychodynamic therapies, descended from Freud, look at patterns that began early in life; they have evidence for several conditions, though less of it and more argument over its quality. Humanistic approaches, from Carl Rogers, put the relationship at the centre. Newer methods include dialectical behaviour therapy for severe emotional difficulty, and EMDR for trauma, which is recommended by guidelines but debated over what part of it does the work. How do you say a therapy "works"? You compare outcomes to a control group and report the gap in standard deviations, the effect size (Cohen's d). By convention 0.2 is small, 0.5 medium and 0.8 large. The picture below turns that into something you can feel. Interactive · What an effect size looks like Small 0.2Medium 0.5Large 0.8 Effect size, d Two normal curves with equal spread. "Chance a treated person does better" is the probability of superiority, Φ(d ÷ √2). Without JavaScript: d = 0.2 means a randomly chosen treated person does better than a random control about 56% of the time; d = 0.5, about 64%; d = 0.8, about 71%. Notice that even a "large" effect leaves wide overlap: many people in the control group do better than the average treated person. That is true of nearly everything in this field and it is not a reason for despair. A Lancet network meta-analysis in 2018 (Andrea Cipriani and colleagues; 522 trials, 116,477 people) found that all 21 antidepressants it studied beat placebo in short-term trials, with odds ratios between about 1.4 and 2.1, so the benefit is real but modest and not the same for everyone. Psychotherapy and medication have each been shown to help in depression, and for many anxiety conditions psychotherapy does as well or better, with longer-lasting benefit once treatment stops. A caution. Published results flatter treatments. In a 2008 analysis in the New England Journal of Medicine, Erick Turner and colleagues compared FDA records with journal articles for antidepressant trials, and found the journal literature looked about 94% positive while the FDA files showed about half. Pre-registration and publishing negative results have helped, but trial size, who is excluded and who drops out still affect how well a result transfers to a real patient. Another persistent finding is that the relationship between patient and therapist predicts outcomes across many methods, which is part of why a good fit matters. Whether all therapies are equally good (the "Dodo bird verdict") is still argued. Global mental health: who gets care Most of the world's research comes from rich countries, and most of the world's people with mental disorders live elsewhere. The World Health Organization reports that only about a third of people with depression receive formal care, and the gap is far wider in poor countries. Its Mental Health Atlas 2020 put the global median at 13 mental health workers per 100,000 people, with fewer than two per 100,000 in low-income countries and over 60 in high-income countries. Interactive · One worker, how many people? Low-incomeGlobal medianHigh-income Mental health workers per 100,000 people The one-in-seven figure is WHO's 2023 estimate; the worker figures are from the 2020 Atlas (the low and high ends as reported in secondary summaries). Not everyone with a disorder needs a specialist, so this is a measure of scale, not of waiting lists. The bar is drawn on a square-root scale. Without JavaScript: at 13 workers per 100,000, about 1,100 people with a disorder (one in seven) per worker; at 60, about 240; at 1.4, over 10,000. One response is to take treatment out of the clinic. The Friendship Bench, created in Harare, Zimbabwe, by the psychiatrist Dixon Chibanda, trains lay health workers, often older women known as "grandmothers", to give six sessions of problem-solving therapy on a bench in the grounds of a clinic. In a 2016 randomised trial in JAMA with 573 adults, six months later about 14% of the Friendship Bench group still had symptoms of depression on the measure used, against about 50% of those who got usual care. It spread to several other countries. The WHO's mhGAP programme and Vikram Patel's work on task-sharing follow the same logic: train more people to deliver a limited, evidence-based set of skills. Culture shapes how distress shows. The DSM now lists "cultural concepts of distress": ataque de nervios, common in Latin America and the Caribbean; kufungisisa ("thinking too much") among the Shona of Zimbabwe, which can describe worry or depression; susto, fright that is believed to cause illness. These do not map neatly onto Western categories, and a good clinician asks about meaning before sorting a person into a diagnosis. What a graduate just sees Most symptoms form a continuum; a diagnosis puts a line through it, so people on either side of it are alike. The manuals disagree at the edges; a label is a tool with error, not a discovery. A positive on any test, for a rare condition, is mostly a false alarm until someone checks. Causes are many and interacting; one cause such as "a chemical" or "a gene" is a simplification. A treatment's effect is an average with wide overlap, so a good result for most is not a promise for you. Published trials look better than the full record; ask what was left out. Most people who need care get none, and trained lay workers can do a lot of good. Distress has local language; ask what it means to the person. Where it shows up next Organizational Behavior (34/53): burnout, workplace mental health and the evidence behind employee wellbeing programmes. Developmental Psychology (35/24): how early adversity and attachment shape later risk. Biological Psychology (35/20) and Personality & Individual Differences (35/26): the brain and the traits that clinical work builds on. Public Economics (34/49): who pays for care, and why prevention is cheaper than it looks. In the world: school counselling, crisis lines, workplace insurance, and the headlines that say a drug or an app "cures" something. If you want to go deeper 📘 Textbook. David Barlow and Mark Durand, Abnormal Psychology: An Integrative Approach (Cengage, many editions), or Ronald Comer, Abnormal Psychology (Worth). Both are common undergraduate texts. 📘 Reference. American Psychiatric Association, DSM-5-TR (2022), and the free online ICD-11 browser at icd.who.int. 📘 Gentler. Allen Frances, Saving Normal (2013), a critical view of diagnostic inflation by the man who chaired the DSM-IV task force, and Susannah Cahalan, The Great Pretender (2019). 🌐 Free course. Open Yale Courses PSYC 110, Introduction to Psychology (Paul Bloom) includes lectures on mental illness and its treatment. The free Noba textbooks have clinical modules. The takeaway Disorders are patterns on a continuum, defined by distress and impairment, and described by two manuals that agree on much and not on all. Causes are biological, psychological and social at once. Several therapies help on average, with real uncertainty, and the largest problem worldwide is not choosing a treatment but getting any. Try this: the next time a headline claims a new therapy or drug works, find three things: the size of the effect, the comparison group, and whether the result was reproduced. Sources & Further Reading Sources. American Psychiatric Association, DSM-5 (18 May 2013) and DSM-5-TR (18 March 2022; includes prolonged grief disorder); WHO, ICD-11 (endorsed by the World Health Assembly on 28 May 2019, in effect 1 January 2022; complex PTSD 6B41, gaming disorder 6C51; gender incongruence in the sexual-health chapter; personality disorder graded by severity). Wakefield, American Psychologist 1992; Rosenhan, Science 1973, and Cahalan 2019; homosexuality removed from the DSM in 1973 and from the ICD in 1990. Regier et al., "DSM-5 Field Trials in the United States and Canada, Part II," American Journal of Psychiatry 2013. WHO fact sheet on mental disorders (2023 figures: about 1.2 billion people, 470 million with anxiety, 322 million with depression; about a third of people with depression receive formal care): who.int/news-room/fact-sheets/detail/mental-disorders. WHO, Mental Health Atlas 2020 (global median 13 workers per 100,000; WHO news release of 8 October 2021); low and high ends as given in secondary summaries and not checked against the report. Engel, Science 1977; Moncrieff et al., Molecular Psychiatry 2022, and the critical comment by Jauhar and colleagues (2023); Cipriani et al., The Lancet 2018 (522 trials, 116,477 participants, odds ratios 1.37 to 2.13); Turner et al., New England Journal of Medicine 2008. Chibanda et al., "Effect of a primary care-based psychological intervention on symptoms of common mental disorders in Zimbabwe," JAMA 2016 (573 participants, 24 clinics; depression symptoms 14% against 50% at six months). Twin heritability of about 80% for schizophrenia and about 40% for major depression: Sullivan and colleagues (2000, 2003), given as round figures. Computed in node from the code on this page: prevalence above a cut-off of 1.5 SD is 6.7%; PPV 15.5% at 2% prevalence and 69.2% at 20%; probability of superiority 55.6%, 63.8% and 71.4% at d = 0.2, 0.5 and 0.8. 📘 Barlow and Durand, Abnormal Psychology 🌐 findahelpline.com 🌐 Open Yale Courses PSYC 110 Step 14. Surgery, Anaesthesia & Emergency Care: The First Hour and the Operating Room ------------------------------------------------------------------------------------- Why: Operations, anaesthesia and the first hour after an emergency. Surgery is a controlled injury done to cure a worse one. What makes it safe is not a steady hand alone, but germs kept out, a sleeping patient watched number by number, a team that checks, and a system that gets the right person to the right room in time. On 16 October 1846, in an operating theatre at Massachusetts General Hospital in Boston, a dentist named William Morton held a flask of ether to the face of a man named Gilbert Abbott while the surgeon John Collins Warren removed a tumour from his neck. The patient did not scream. Warren is reported to have told the crowded gallery, "Gentlemen, this is no humbug." The room is still called the Ether Dome. Within months the news had crossed the Atlantic. Today the idea that an operation should be painless is so obvious that we forget that, for thousands of years, speed was the surgeon's chief skill. This piece explains what surgery and emergency care do and why; it is not medical advice and not a guide to doing anything. If someone needs help, call your local emergency number. The course in one box Who takes itMedical students in the clinical years (a surgery rotation, an anaesthesia attachment and emergency medicine are required almost everywhere), then residents in surgery, anaesthesia and emergency medicine. Nurses, paramedics and operating-department practitioners learn it too. What it is forTreating what drugs cannot fix, keeping people alive and pain-free while it is done, and rescuing people in the first minutes of an injury or collapse. Comes afterAnatomy (35/54) and Pharmacology (35/56). Leads toSurgical, anaesthetic and emergency careers, and links back to Medical Ethics & Law (35/60) for consent in a hurry and Public & Global Health (35/61) for access. What this piece gives youYou will not be able to operate, give an anaesthetic or resuscitate anyone; that takes years and certified courses. You will see why surgery is a bet, why washing hands mattered so much, why the oxygen number on the clip is read the way it is, what a checklist really does, why the "golden hour" is contested, and how triage and global access work. Easier first reads: Pain Explained (01/06) and Your Immune System, Day by Day (01/10). What surgery does, and what it costs Surgery does five things: it removes (an inflamed appendix, a tumour), repairs (a broken bone, a torn artery), reconstructs, replaces (a joint, a kidney) or diverts (a bypass). All of them are a trade. The operation brings a certain, front-loaded risk (bleeding, infection, the anaesthetic) to avoid a risk that is spread over time (the disease). Whether it is worth it depends on how big each risk is and on what the person values. Surgeons call the choice of when, and whether, to operate indications, and some of the best surgical judgment is knowing when not to cut. risk of waiting over n years = 1 − (1 − r)nr = yearly chance of serious harm if nothing is done; compare with the operation's one-off risk Interactive · Operate or wait? (a fictional condition) Yearly risk of harm if left alone (%) Risk of the operation (%, one-off) Years you look ahead Fictional condition with round numbers; real decisions depend on the person, the operation and alternatives. Not a decision tool. With a 2% yearly risk and a 3% operation risk, waiting has an 18.3% chance of harm by year 10, far above 3%, and the lines cross at 1.5 years. Drop the yearly risk to 0.3% and the lines cross after 10 years: for older or frailer people, a slow disease may never outrun the operation. Asepsis and anaesthesia: the two revolutions Before the 1840s, operations were fast because patients were awake; before the 1860s, many who survived the operation died of infection afterwards. Two advances changed both. Anaesthesia. Not all of it began in Boston. Hanaoka Seishū in Japan operated on a 60-year-old woman for breast cancer on 13 October 1804 using a herbal mixture, tsūsensan. Crawford Long in Georgia removed a neck tumour under ether in 1842 but did not publish until 1849. Horace Wells tried nitrous oxide in 1844; his public demonstration in Boston failed. Morton's 1846 success was public and reproducible, and in 1847 the Edinburgh obstetrician James Young Simpson used chloroform. In 1853, John Snow, the same Snow of the pump, gave chloroform to Queen Victoria for the birth of her eighth child. Asepsis. In 1847, at the Vienna General Hospital, Ignaz Semmelweis noticed that mothers delivered by doctors (who also did autopsies) died of childbed fever about 10% of the time, against under 4% in the midwives' clinic. He ordered handwashing in chlorinated lime, and deaths in the doctors' clinic fell from 18.3% in April to 2.2% in June. The common telling, a lone genius who was ignored and went mad, is too simple. He had no theory, published little for years, and the "cadaverous particles" idea was one of many; he died in 1865 in an asylum, from an infected wound, and the cause of his breakdown is still debated. In 1867 Joseph Lister, building on Pasteur's germ theory, published his method of treating wounds with carbolic acid. Antisepsis kills germs in the wound; asepsis, which followed (steam-sterilised instruments, gowns, and by about 1890 rubber gloves), keeps them out in the first place. Together with anaesthesia, they made long, careful operations possible. A modern anaesthetic is not just "sleep". It aims at three things (the classic triad): unconsciousness, pain relief and muscle relaxation, each given by different drugs and each watched, with the anaesthetist's job being to keep the patient alive and stable while the surgeon works. Watching the numbers: oxygen and the pulse oximeter During an operation the patient cannot say how they feel, so numbers speak for them: heart rhythm, blood pressure, breathing, carbon dioxide breathed out, temperature, and oxygen. The oxygen number comes from the pulse oximeter, the clip on a finger. It shines red (660 nm) and infrared (940 nm) light through the tissue; blood rich in oxygen absorbs the two colours differently from blood poor in it, and the device reads the ratio at each pulse. Takuo Aoyagi and Michio Kishi built the first one at Nihon Kohden in 1972, and by 1987 it was part of the US standard of care for general anaesthesia. The reading is a percentage saturation (SpO2), but what matters is how much oxygen reaches the tissues. Haemoglobin loads oxygen along an S-shaped curve: the top of the curve is flat, the bottom is a cliff. oxygen content ≈ 1.34 × Hb × saturation + 0.003 × PO2mL of O₂ per dL of blood · Hb in g/dL · delivery = content × cardiac output Interactive · The flat top and the cliff Oxygen pressure in blood (mmHg) Haemoglobin (g/dL) Hill curve with n = 2.7 and P50 = 26.8 mmHg, a standard textbook approximation; delivery assumes a cardiac output of 5 L/min. Not a clinical calculator. At a pressure of 100 mmHg and haemoglobin 15 the saturation is 97% and about 990 mL of oxygen are delivered each minute. Fall to 60 mmHg and saturation is still 90%; fall to 40 and it is 75%. Because the top is flat, the reading hardly moves at first and then drops fast. Drop haemoglobin to 8 and saturation still reads 97% while delivery falls to about 540 mL a minute: the number on the clip is not the whole story. Pulse oximeters also have known limits: they can over-read in people with darker skin, be fooled by carbon monoxide, and fail with movement or poor circulation. The checklist and the Swiss cheese Most surgical harm is not a dramatic slip but a gap: the wrong side marked, an antibiotic not given, a swab left behind, an instrument missing. The psychologist James Reason's "Swiss cheese" picture (1990) describes this: every defence has holes; harm happens only when the holes in several slices line up. In 2008 WHO launched a Surgical Safety Checklist of about 19 items, read aloud at three pauses (before anaesthesia, before the first cut, before leaving the room). In 2009 Alex Haynes and colleagues published a study in the New England Journal of Medicine across eight hospitals in eight cities (Toronto, New Delhi, Amman, Auckland, Manila, Ifakara, London, Seattle). After the checklist, major complications among inpatients fell from 11.0% to 7.0% and deaths from 1.5% to 0.8%. It was a before-and-after study with no control group, and later work was mixed: a 2014 study of Ontario hospitals found no significant change after a checklist was required. Evidence suggests that how a checklist is used, with real pauses and a team willing to speak up, matters more than the paper. Interactive · Layers of defence Errors per 1,000 operations Checking layers Each layer catches (%) Errors every layer misses (%) Illustrative arithmetic, not data. Residual = errors × (shared blind spot + (1 − blind spot) × (1 − catch rate)layers). With 10 errors in 1,000 operations and three layers that each catch 60%, only 0.64 get through if the layers are independent. But if 10% of errors are of a kind every layer misses (the same blind spot, the same tired people), the figure is 1.58. More layers do little once the shared blind spot dominates, which is the real argument for a culture where anyone may stop the line. Trauma and the golden hour Injury is a leading cause of death for the young worldwide, and care begins before the hospital. A trauma system links the roadside, ambulance, a hospital with the right resources, and rehabilitation. One idea has dominated: the golden hour, which says that the first hour after major injury decides survival. R Adams Cowley, who built the shock-trauma centre in Baltimore, popularised it, and it has shaped ambulance targets worldwide. The evidence is mixed, and that is the honest account. A 2001 review by Lerner and Moscati, "The golden hour: scientific fact or medical urban legend?", found no scientific basis for a magic 60 minutes. A 2010 US cohort of 3,656 injured people (Newgard and colleagues) found no association between ambulance times and death, though critics note most patients were reached in under an hour, so it could not test the idea well. And in the US military, after a 2009 order to evacuate critically injured casualties within 60 minutes, median transport time in Afghanistan fell from 90 to 43 minutes and the share reaching a hospital within the hour rose from 24.8% to 75.2%; the study (Kotwal and colleagues, 2015) reported fewer deaths, partly from a modelled projection. The usual reading is that time is critical for people who are bleeding fast, and a clock is a poor stand-in for the condition: get the bleeding stopped, get the person to definitive care. Interactive · The bleeding clock Body weight (kg) Blood loss rate (mL/min) Blood volume taken as 70 mL per kg of an adult. Doctors commonly grade loss as under about 15%, 15 to 30%, 30 to 40%, and over 40% of volume. A teaching illustration with no bearing on any real injury; the body also compensates and stops some bleeding itself. For a 70 kg adult with about 4.9 litres of blood, a loss of 100 mL a minute reaches 15% in under 8 minutes and 40% in about 20. At 1,000 mL a minute, it is two minutes. At 10 mL a minute it takes more than 3 hours. The same "hour" can be far too long or comfortably long; that is the contested part. Triage, CPR and the first minutes Triage, from the French trier, to sort, means treating people in order of need and chance to benefit, not arrival. Baron Dominique-Jean Larrey, Napoleon's chief surgeon, argued for it on the battlefield and introduced his "flying ambulances". Emergency departments use five-level scales such as the Emergency Severity Index (developed in 1998); in disasters, simple systems such as START (Newport Beach, 1983) sort casualties into immediate (red), delayed (yellow), minor (green) and expectant (black). Triage is imperfect; reviews report under-triage in a third of some hospital cohorts, and critics say colour tags have not been shown to predict survival accurately. Interactive · A day in the waiting room Patients arriving per hour Clinicians working Simulation of 600 invented patients in levels 1 (most urgent) to 5, mean times 30, 25, 20, 12 and 8 minutes per patient. The same patients are used in both queues. Not a real hospital. With the starting settings the load is about 85%. In a first-come queue everyone waits about 35 minutes; under triage level 1 patients are seen in about 7 minutes, and level 5 patients wait about two hours. That is the price, and the point. The first minutes of a cardiac arrest, when the heart stops pumping, follow the same logic. At awareness level, current guidance from the main resuscitation councils (ILCOR, the American Heart Association, the European Resuscitation Council) asks a bystander to do three things: call the emergency number, push hard and fast in the centre of the chest, and use an automated external defibrillator (AED) as soon as one arrives, following its voice instructions. An AED analyses the heart rhythm itself and only advises a shock when one is needed. The usual target is 100 to 120 compressions a minute, a little over 5 cm deep in an adult. Survival from cardiac arrest outside hospital is low: one US registry summary from 2018 found about 10% survived to discharge overall, and about 35% when a bystander used an AED. Please take a certified first-aid course; this page is not training. Interactive · Feel the tempo Tap here steadily Reset Tap at least three times. Measures only how steady your tapping is on this device. It is not CPR training and has nothing to do with depth or technique. Surgery for everyone: global access Surgery is not a luxury of rich countries; it is a part of care for injury, childbirth, appendicitis, cleft lip and cataract. But the Lancet Commission on Global Surgery (2015) estimated that five billion people lacked access to safe, affordable surgical and anaesthesia care when needed, and that 143 million additional procedures a year were needed in low- and middle-income countries. The shortages are of people (surgeons, anaesthetists, nurses), equipment (oxygen, working pulse oximeters), and money, since a family may be bankrupted by an operation. Several answers have been tried. Charities such as Lifebox have supplied pulse oximeters to operating theatres that lacked them. Task-sharing lets trained non-physician clinicians do some procedures, as with surgical technicians who perform caesarean sections in parts of Mozambique. The WHO checklist is free. None of these replaces the main need: reliable oxygen, electricity, blood and trained teams, and health systems built so that surgery is part of the plan (see Public & Global Health, 35/61). What a graduate just sees An operation is a bet: a certain, front-loaded risk against a spread-out one. Infection is controlled by keeping germs out, not just killing them in the wound. Under anaesthesia the monitors are the patient's voice; read each number for what it measures and what it misses. The oxygen curve is flat at the top and steep at the bottom, and haemoglobin matters as much as saturation. Safety is layers with holes; checklists work when the team actually pauses and speaks up. For trauma, ask about the bleeding rate, not the clock alone. Triage trades fairness-by-arrival for fairness-by-need; someone always waits longer. For a cardiac arrest: call, push, shock, and learn it properly beforehand. Most people on Earth cannot get a safe operation when they need one. Where it shows up next Anatomy (35/54) and Pharmacology (35/56) sit under every step: where to cut, how drugs act. Physiology (35/49): blood volume, oxygen delivery and shock. Medical Ethics & Law (35/60): consent for emergencies, and who may be treated first. Public & Global Health (35/61): injury prevention, ambulance systems, and surgical access. In the world: operating theatres, ambulances and helicopters, emergency departments, disaster response, and every shopping-centre AED. If you want to go deeper 📘 Textbook. Bailey & Love's Short Practice of Surgery (CRC Press). A standard UK and Commonwealth undergraduate text. 📘 Textbook. Sabiston Textbook of Surgery (Elsevier), and for anaesthesia Miller's Anesthesia (Elsevier), the reference works. 📘 Gentler. Atul Gawande, Complications (2002) and The Checklist Manifesto (2009); Lindsey Fitzharris, The Butchering Art (2017), the story of Lister. 🌐 Free resources. WHO's Safe Surgery pages and the free courses at OpenWHO. For first aid, take a certified course from the Red Cross or a national resuscitation council. The takeaway Surgery became safe through four things done in order: anaesthesia (1846), germ control (1860s onward), monitoring (above all oxygen) and teamwork with checks. Emergency care adds a fifth: getting the right care to the right person fast. The arithmetic is gentle (risk of waiting, the oxygen curve, layers of defence, a bleeding clock), and the biggest unsolved problem is access. Try this: this week, find the nearest AED in a place you visit often (a station, a shop, a gym). Note where it is and what the sign looks like. And look up when a certified first-aid course is next offered near you. (This is awareness, not training.) Sources & Further Reading Sources. Anaesthesia history: Hanaoka 1804, Long 1842, Wells 1844, Morton 16 October 1846, Simpson 1847, Snow 1853 (encyclopaedia summary of the history of general anaesthesia). Semmelweis: encyclopaedia summary (Vienna 1847; mortality 18.3% to 2.2%; died 1865). Lister 1867 and rubber gloves c.1890 as standard history. Pulse oximetry: encyclopaedia summary (1972, 1987). Hill curve: standard textbook approximation. WHO checklist: Haynes et al., N Engl J Med 2009 (PubMed): deaths 1.5% to 0.8%, complications 11.0% to 7.0%; Ontario study (2014) from memory. Golden hour: Lerner and Moscati (2001); Newgard et al. (Ann Emerg Med 2010); Kotwal et al. (JAMA Surgery, 2015). Triage: Larrey, START (1983), ESI (1998); under-triage figure from an encyclopaedia summary of a 2014 review. CPR: ILCOR/AHA/ERC emphasis, rate and depth from an encyclopaedia summary of the guidelines (check the latest edition). Global surgery: Lancet Commission on Global Surgery (Meara et al., 2015), figures as reported in summaries. Interactive numbers computed in node. Not medical advice. 🌐 WHO: Safe Surgery 🌐 OpenWHO 📘 Gawande, The Checklist Manifesto 📘 Fitzharris, The Butchering Art Step 15. Public & Global Health: Health Beyond the Clinic --------------------------------------------------------- Why: Health beyond the clinic: populations, systems and the world. Medicine treats the person in front of it; public health asks why that person is the one in the chair, and what would have kept the next thousand out of it. Its tools are counting, water, laws, money and vaccines. In August 1854, cholera hit Broad Street in Soho, London. The doctor John Snow marked the deaths on a map, saw them cluster around one water pump, and persuaded officials to remove its handle. He did not know about bacteria; he had a map, a hunch and a count. That is public health in one picture: no cure, no drug, a change made to a shared thing, and a drop in deaths among people who never knew they had been at risk. A hundred and seventy years later the shared things are tobacco taxes, vaccination schedules, clean air, seat belts and how a country pays for its hospitals. This piece explains how public health works; it is not medical advice. For a health question, see a clinician; in an emergency, call your local emergency number. The course in one box Who takes itMedical students (often as "community medicine" or "population health", in the later clinical years), nursing and public-health students, and anyone in health policy. Master of Public Health degrees draw on epidemiology, biostatistics, health policy, environmental health and social science. What it is forPreventing disease and prolonging life across whole populations, and deciding what a society should spend its health money on. Comes afterEpidemiology & Evidence-Based Medicine (35/58), for how causes are found, and Public Policy (35/16), for how problems become policies. Leads toHealth-policy and global-health careers; and it connects back to Medical Ethics & Law (35/60), where rationing and mandates are argued. What this piece gives youYou will not be able to design a vaccination programme. You will see why most disease is made outside hospitals, why the mix of diseases changes as countries grow richer, how health is counted, the four ways health systems are paid for, what herd immunity really is, and what tobacco and pandemics taught the field. Easier first reads: Your Immune System, Day by Day (01/10) and Where Public Money Goes (18/08). Health is made outside the clinic Ask what keeps a population healthy and doctors are, to the surprise of many, only part of the answer. Clean water and sewers, food, housing, work, education, income and the air all matter more for most diseases than treatment. A famous argument by the historian Thomas McKeown (1976) said that Britain's death rate fell in the nineteenth century mainly because people ate better, not because medicine improved. Later historians such as Simon Szreter credited public works and sanitation as well. The two sides disagree about the mix, not about the point: a society's health is mostly made upstream. The same is true today. In the Whitehall studies, led by Michael Marmot from the late 1960s, British civil servants, all with secure jobs and access to the same health service, showed a gradient: each step down the grade ladder came with worse health. The WHO's Commission on Social Determinants of Health (2008) put the lesson in its title, Closing the Gap in a Generation. Social position, not only germs and genes, gets under the skin. A second lesson comes from the epidemiologist Geoffrey Rose (1981, 1985): a population approach shifts everyone a little; a high-risk approach finds and treats the people at the top of the distribution. The high-risk approach is efficient for each person treated, but most cases come from the large middle group, because there are so many more people there. Rose called the result the prevention paradox: a measure that brings large benefits to the community offers little to each individual. Interactive · The prevention paradox Risk doubles every (score units) Shift achieved (score units lower) High-risk group (top % of scores) Illustrative model, not real data: 100,000 people with a bell-shaped risk score (mean 100, spread 15); yearly risk is 0.5% at a score of 100 and doubles every set number of units. "Population" lowers everyone's score by the shift; "High-risk" lowers only the top group's. With the starting values there are about 858 cases a year among 100,000 people. The top 10% of scores produce only 41% of them, so 59% of cases occur in people nobody would call high-risk. Shifting everyone by 5 units prevents 29% of cases but needs about 398 person-years of effort per case prevented; treating only the top 10% prevents 12% but needs about 99. The narrow approach is four times as efficient and prevents less than half as much. The transition from infection to chronic disease As countries get richer, the diseases that kill them change. Abdel Omran named this the epidemiological transition in 1971: first an age of pestilence and famine, then of receding pandemics as sanitation and food improve, then of degenerative and man-made diseases such as heart disease, cancer and diabetes. Later writers added stages, and critics point out that it does not run in a straight line: many low- and middle-income countries carry a double burden of infections and chronic disease together, and some "chronic" diseases turn out to have infectious causes. The latest World Health Organization data (for 2023) show it clearly. About 61 million people died that year; the top causes worldwide were ischaemic heart disease (9.5 million), stroke (6.8 million), chronic obstructive pulmonary disease (3.6 million), lower respiratory infections (2.7 million), Alzheimer's and other dementias (2.1 million), diabetes (2.0 million) and lung cancers (1.8 million). WhereWhat the top ten causes of death look like, WHO 2023 Low-income countriesSeven of ten are communicable. Lower respiratory infections lead; diarrhoeal diseases are fifth; malaria, tuberculosis and HIV/AIDS appear. High-income countriesNine of ten are noncommunicable. Heart disease and stroke are in the top three; dementia is third and lung cancer fourth; lower respiratory infections (sixth) are the only infection. Counting the burden: deaths, years and DALYs Deaths alone mislead: a death at 30 costs more years than one at 90, and a disease that disables without killing does not appear. The disability-adjusted life year (DALY) counts one lost year of healthy life, whether lost by early death or by living with illness. It is the main unit of the Global Burden of Disease (GBD) study, run by the Institute for Health Metrics and Evaluation (IHME) in Seattle, and WHO publishes its own estimates; the two differ in methods and sometimes in numbers. DALY = YLL + YLDyears of life lost to early death + years lived with disability (cases × weight × duration) A disability weight runs from 0 (full health) to 1 (equivalent to death) and comes from surveys of how people rank states of health. The weights are debated: the GBD revised many between 2004 and 2010 (blindness fell from 0.594 to 0.195), dropped the early age-weighting that valued young adults' years more, and dropped discounting of future years. Critics say any such weighting hides value judgments; defenders say that without a common unit, we cannot compare a heart attack with a lifelong back problem. Interactive · Which condition costs more? Sudden and fatalChronic and mildDisabling for life Deaths Years of life lost per death People living with it Disability weight Years lived with it Invented conditions with round numbers. The presets are not real diseases. The first preset gives 1,000 deaths × 30 years = 30,000 years of life lost, and no disability. The second has no deaths at all but 50,000 people living for 10 years at weight 0.2: 100,000 years lived with disability. By deaths alone the first looks worse; by DALYs the quiet, chronic one is more than three times as big. That is why lower back pain, depression and hearing loss rank high in burden tables even though they rarely kill. Who pays: four ways to run a health system Every health system answers the same question: who pays, and who provides? Four patterns recur, and most countries mix them. Beveridge model (named after William Beveridge's 1942 report): tax-funded, government-run. The UK's National Health Service, founded in 1948, is the classic example; Spain and the Nordic countries are close. Bismarck model: insurance funds paid by employers and workers, with private providers. It began with Germany's sickness insurance law of 1883; Germany, France and Japan work this way in different forms. National health insurance: one public insurer paying private providers. Canada's provinces and South Korea use it. Out-of-pocket: patients pay when they are sick. It is the main route where public systems are weak, and it pushes families into poverty. Real systems vary: Switzerland requires everyone to buy regulated private insurance; the United States combines employer plans, public programmes (Medicare, Medicaid) and the uninsured; Brazil's tax-funded Sistema Único de Saúde was set up by the 1988 constitution; Thailand introduced a tax-funded universal coverage scheme in 2002; Rwanda built community insurance funds with donor money. Interactive · Spending around the world % of GDPPer personOut-of-pocket % Data: WHO Global Health Expenditure Database as republished by Our World in Data, retrieved 1 November 2026; GDP share is for 2023, per-person spending (international dollars, adjusted for prices) and out-of-pocket share for 2023 or 2024 depending on the country. "Out-of-pocket" is the share of all health spending paid directly by patients. System labels are simplified. The United States spent 16.7% of its GDP on health in 2023, about $13,500 per person; Germany 11.7%, the UK 11.0%, Japan 10.7%. In Nigeria 72% of health spending was out of pocket, in India 44%, in Mexico 41%; in Thailand, which spends only 4.5% of GDP, it was about 10%. Spending, coverage and health are three different things. Vaccines, herd immunity and eradication The central number of infectious-disease dynamics is R0, the average number of people one case infects in a fully susceptible population. If R0 is above 1, an outbreak grows. Immunity shrinks the number: when a fraction v of people are immune, each case infects R0(1 − v) others, and the outbreak dies when that falls below 1. The needed fraction is the herd immunity threshold. threshold = 1 − 1 ÷ R0divided by vaccine effectiveness if it is below 100%; assumes people mix randomly The usual figures are 71 to 83% for smallpox, 80 to 86% for polio and 92 to 94% for measles (R0 about 12 to 18). Real populations mix unevenly, so pockets of low coverage can keep a disease alive even when the national average looks high. Interactive · Run an epidemic R0 of the disease Vaccinated (%) Vaccine effectiveness (%) A simple model (SIR) with everyone mixing equally, an infectious period of 5 days, and 1 in 10,000 infected at the start. The vaccine is "all-or-nothing" on a fraction of people. Real epidemics are messier. Set R0 to 2.5 with no vaccination: the model gives a peak of about 24% of people infected at once and about 89% infected in total. Vaccinate 60% with a 95%-effective vaccine and 57% are immune, just under the 60% threshold: the effective R is 1.08 and only about 5% are ever infected. At 70% coverage, 66% are immune, the effective R is 0.84 and the outbreak barely starts. For measles the coverage needed is above 90%, which is why a few unvaccinated schools can bring it back. Eradication is the extreme case: drive the number to zero everywhere. Smallpox is the only human disease so far. The WHO began an intensified programme in 1967, the last natural case was diagnosed in October 1977 (Ali Maow Maalin, in Somalia), and eradication was certified in May 1980; the last death was a laboratory accident in Birmingham in 1978. Stocks of the virus still exist at two WHO-approved laboratories, in the United States and Russia. Rinderpest, a cattle disease, was declared eradicated in 2011. Polio is near: the global programme began in 1988, wild poliovirus types 2 and 3 were declared eradicated in 2015 and 2019, and the African region was certified free of the wild virus in 2020. Type 1 persists in Pakistan and Afghanistan: 99 cases in 2024 (74 and 25), and about 38 reported by October 2025 (the full-year count is not verified here). The oral vaccine can also, rarely, seed outbreaks in under-vaccinated communities. Tobacco: a long fight with a known cause Smoking is the great proof that prevention works slowly. In 1950 Richard Doll and Austin Bradford Hill published a case-control study linking cigarettes to lung cancer, followed by their British doctors' cohort in 1954 (see Epidemiology & Evidence-Based Medicine, 35/58). The industry's response, documented in later litigation and in books such as Merchants of Doubt, was to manufacture doubt. Today WHO says tobacco kills more than 7 million people a year, of whom over 1.6 million are non-smokers exposed to smoke (2023 data). The response was the first treaty negotiated under WHO: the Framework Convention on Tobacco Control, adopted in 2003 and in force since 2005, with 183 parties by WHO's count. It is backed by the "MPOWER" package (2007): monitor use, protect people from smoke, offer help to quit, warn about dangers, enforce bans on advertising, raise taxes. Raising tobacco taxes is widely regarded as one of the most effective measures. Australia introduced plain packaging in 2012; others followed. Newer products such as e-cigarettes divide experts: some health bodies see them as a route off cigarettes, others as a route on to nicotine for young people, and the evidence is still developing. Pandemics and preparedness An epidemic is more cases than expected in a place; a pandemic is one that spreads across continents; a disease is endemic when it circulates steadily. The 1918 influenza is the benchmark, with estimates of deaths spanning a wide range, commonly put near 50 million. HIV/AIDS is a pandemic that is still going; the 2014 to 2016 Ebola outbreak in West Africa caused more than 11,000 deaths; the 2009 H1N1 influenza spread worldwide. COVID-19 showed how hard it is to count a pandemic. WHO estimated 14.9 million excess deaths (range 13.3 to 16.6 million) in 2020 and 2021, far more than officially recorded; an IHME-led study published in 2022 estimated more than 18 million. The difference comes from different methods and the fact that many countries record few deaths at home. Lessons the field drew included early surveillance, fast testing, supply stocks, trust and clear communication, and fair access to vaccines, which was unequal between rich and poor countries. The International Health Regulations (2005) require countries to report events; in May 2025 the World Health Assembly adopted a Pandemic Agreement, though as far as I could verify its annex on sharing pathogens and benefits was still being negotiated. What a graduate just sees Most health is made upstream (water, food, housing, income); a clinic treats what upstream failed to prevent. A small shift across a whole population can prevent more cases than a large shift in the few at highest risk. Deaths undercount burden; years lost and years lived with disability change the ranking. As countries grow richer the leading causes of death move from infections to chronic disease, often leaving a double burden in between. Health systems differ in who pays and who provides; spending, coverage and outcomes are three different things. Herd immunity is 1 − 1/R0, and it protects those who cannot be vaccinated only while coverage stays high everywhere. Every statistic about a pandemic depends on who counted and how; ask for the method. The best public-health measures are often boring: taxes, laws, pipes and schedules. Where it shows up next Epidemiology & Evidence-Based Medicine (35/58) supplies the methods (cohorts, trials, bias) that public-health claims rest on. Public Policy (35/16) and Political Economy (35/17): who pays, and why reforms stall. Medical Ethics & Law (35/60): rationing, mandates and consent at population scale. Microbiology (35/50) and Immunology (35/57): the biology behind R0, vaccines and resistance. In the world: ministries of health, WHO and national disease-control agencies, NGOs and foundations, health insurers, and every school-vaccination schedule. If you want to go deeper 📘 Textbook. Michael Marmot, The Health Gap (2015) for the social-determinants view, and a standard introduction such as Bernard Turnock's Public Health: What It Is and How It Works (Jones & Bartlett), widely used in public-health schools. 📘 Textbook. Detels et al., Oxford Textbook of Global Public Health (Oxford University Press). The reference work for the global picture. 📘 Gentler. Hans Rosling, Factfulness (2018), and Steven Johnson, The Ghost Map (2006), the story of John Snow and the Broad Street pump. 🌐 Free course and data. OpenWHO hosts free WHO courses; Our World in Data's health pages and the IHME GBD Compare tool let you explore the numbers. The takeaway Public health is medicine's other half: counting who falls ill and why, then changing the shared conditions (water, laws, prices, vaccines, money) so fewer do. Its arithmetic is simple: a threshold of 1 − 1/R0, a DALY as years lost plus years lived with disability, and a paradox that the biggest gains go to people who never knew they were at risk. Try this: this week, pick one thing near you that protects public health without you noticing (tap water, a food-safety rule, a vaccination record, a smoke-free sign). Find who decided it and when. (This is curiosity, not medical advice.) Sources & Further Reading Sources. Causes of death 2023: WHO, "The top 10 causes of death" fact sheet. Health spending: WHO Global Health Expenditure Database via Our World in Data (retrieved 1 November 2026). DALY definition and weights: encyclopaedia summaries of the Global Burden of Disease method. Omran (1971, Milbank Memorial Fund Quarterly) and later criticism. Herd immunity thresholds: encyclopaedia summary of standard figures. Smallpox: WHO; polio: GPEI, WHO and news reports (2024 and 2025 counts differ by source and date). Tobacco: WHO tobacco fact sheet (2023 data; FCTC adopted 2003, in force 2005, 183 parties). COVID excess deaths: WHO, 5 May 2022 (14.9 million, range 13.3 to 16.6); IHME estimate from memory of the 2022 Lancet paper. Snow, McKeown, Whitehall, Beveridge (1942), Bismarck (1883) and Rose (1981, 1985) as standard history. Interactive numbers computed in node. Not medical advice. 🌐 OpenWHO (free courses) 🌐 Our World in Data: Health 📘 Rosling, Factfulness 📘 Johnson, The Ghost Map Step 16. Biological Psychology: The Wet Machine Behind the Mind --------------------------------------------------------------- Why: The nervous system and the brain. Every thought, mood and night's sleep is made of cells that fire or stay quiet, chemicals that drift across a gap twenty nanometres wide, and blood flow that scanners can see. The course teaches what that does and does not explain. .chips button{min-width:0} .row2{display:flex;gap:8px;flex-wrap:wrap} .row2 button{flex:1 1 90px;min-width:0} .legend{font-family:-apple-system,sans-serif;font-size:.8rem;color:var(--muted)} .brainmap circle{cursor:pointer;stroke:var(--ink);stroke-width:1} Strike a funny bone and your arm jerks before you decide anything. Drink a coffee at four in the afternoon and you lie awake at midnight. A blow to the head changes a person's memory or temper. Someone stops taking a medicine and their mood shifts. In each case something happened in tissue, and something happened in experience, and the two cannot be pulled apart. Biological psychology is the course where psychologists meet that fact. Its job is not to explain the mind by pointing at the brain. It is to learn what the tissue does, in what units, at what speed, and, just as important, which popular stories about "the chemical for love" or "the part of the brain for fear" are too neat to be true. The course in one box Who takes itEvery psychology major, usually in year 1 or 2. Neuroscience, nursing, medicine, biology and pharmacy students take a deeper cousin. The British Psychological Society lists biological psychology among its core areas, and the APA guidelines expect undergraduates to link behaviour to its biological basis. What it is forConnecting what people do and feel to the signalling of nerve cells, hormones, and the body that carries them. Comes afterNo psychology prerequisite; school biology helps. Leads toSensation & Perception (35/21), then Cognitive Psychology (35/22), and Clinical Psychology (35/27), where the chemistry of disorders and drugs comes back. What this piece gives youYou will not be able to read a brain scan. You will see how a neuron decides, why "dopamine is the pleasure chemical" is wrong, what a brain image can claim, and how sleep and learning work. Easier first reads: Why You Feel Before You Think (01/01), Sleep Is a Skill (01/04) and Dopamine Is Not the Pleasure Chemical (01/08). This one goes further. Descriptive only, not medical advice. The neuron: a cell that fires or stays quiet The human brain holds about 86 billion neurons, according to a careful 2009 count that replaced the round "100 billion", along with a roughly equal number of supporting cells. A neuron is a cell that collects inputs and, when enough arrive, sends a brief electrical pulse down its fibre. At rest its inside sits about 70 millivolts below the outside. If inputs push it up to about −55 mV, the threshold, channels open, sodium rushes in, and the voltage swings to roughly +30 mV for about a millisecond before resetting. That pulse is the action potential. Alan Hodgkin and Andrew Huxley worked out its mechanism in 1952 on the giant axon of a squid, and shared a Nobel Prize for it in 1963. Two facts matter more than the details. First, it is all or none: a neuron cannot fire a weak spike. A stronger input does not make a bigger pulse, it makes pulses come more often. The rate is the message. Second, the spike travels along the fibre, and fatty myelin wrapped around it makes the signal jump, so speeds reach roughly 100 metres per second in the fastest fibres, against about a metre per second in thin unwrapped ones. C dV/dt = −(V − Vrest)/R + I, and fire when V reaches thresholdthe leaky integrate-and-fire neuron: input charges it, it leaks back, and a spike resets it Interactive · One neuron deciding Steady input (voltage it would reach) A textbook model, not a real cell: time constant 20 ms, rest −70 mV, threshold −55 mV, 2 ms refractory period. Input is the voltage above rest that the cell would reach if it never fired. Below 15 mV of drive, nothing happens, however long you wait. At 20 mV the cell fires about 34 times a second; at 40, nearly 90. The size of every spike is identical. This is the brain's coding trick: not louder, but faster. The synapse: chemical messages, and the myths about them Neurons do not touch. They are separated by a gap about 20 nanometres wide, the synapse, and the signal crosses it as chemistry. Otto Loewi showed in 1921, with two frog hearts, that a nerve releases a substance rather than just a current. When a spike arrives, the cell releases a neurotransmitter, which binds receptors on the next cell and nudges it toward firing (excitation) or away from it (inhibition). Then the leftover is cleared: pumped back (reuptake) or broken down. In the brain, glutamate is the main excitatory transmitter and GABA the main inhibitory one. Pop psychology turns each transmitter into a mood: dopamine is pleasure, serotonin is happiness, oxytocin is love. The truth is messier. Dopamine neurons fire strongly for unexpected rewards and cues, a "better than expected" signal, which is why they matter for learning and wanting, and not simply for enjoying. Most of the body's serotonin sits in the gut, not the brain. Oxytocin does affect social behaviour, but results from squirting it up the nose have been inconsistent, and many early findings shrank as samples grew. One transmitter plays many roles in different circuits; one behaviour uses many transmitters. Interactive · Release, clearing and a reuptake blocker Normal clearing Reuptake blocked (SSRI-style) Presynaptic firing rate Reuptake activity A toy model in arbitrary units: each spike adds one unit; the gap clears at a rate of 10 + 50 × reuptake per second. With full reuptake, a 10 Hz neuron keeps the gap at about 0.16 units; block 80% of reuptake and the level triples to 0.5. That is, roughly, what drugs like SSRIs do within hours. Yet their effect on mood takes weeks, and the slogan "depression is a chemical imbalance" is a simplification that psychiatrists themselves avoid. A 2022 review argued there is no consistent evidence that low serotonin causes depression; others replied that this was never the standard claim and that the drugs still beat placebo on average (a large 2018 comparison of 21 antidepressants found all outperformed placebo, with modest average effects). What can be said: transmitters matter, the story is not one chemical, and no one should change a medicine because of a diagram. Talk to a prescriber. Hormones are the slow cousin. Glands release them into the blood and they act over minutes to days: cortisol, from a chain called the HPA axis, prepares the body for demands; adrenaline acts in seconds. A nerve impulse reaches your foot in a few hundredths of a second; a hormone takes minutes. Two systems, two timescales. The brain's map: what parts do, and what they do not Where things happen in the brain is one of the oldest questions. In 1848 railway foreman Phineas Gage survived an iron bar through the front of his skull, and accounts say his behaviour changed (the details were exaggerated in retelling). In 1861 Paul Broca examined a man who could understand speech but say almost only one syllable; the damage was in the left frontal lobe. In 1953 surgeons removed parts of Henry Molaison's temporal lobes to treat epilepsy; he could no longer form new lasting memories of events, yet he could still learn a new motor skill. Cases like these show that some functions depend on particular tissue. Do not conclude that each part has one job. Functions are networks: damage to one place can break a skill that many places share. Three myths to drop. We do not use 10% of the brain; imaging shows activity throughout, and the brain, about 2% of body weight, uses around a fifth of the body's resting energy. "Left-brained" and "right-brained" people do not exist; a 2013 analysis of over 1,000 scans found no person-wide dominance, although the two halves do specialise a little (language is usually left). And the neuron count is lopsided: the cerebellum, tucked under the back of the brain and often thought of as just for coordination, holds roughly 80% of the neurons. Interactive · If this part is damaged… A cartoon, not an atlas. Pick a region or tap its circle. Deficits are textbook summaries of famous cases and vary by person. Brain imaging: what a scan can and cannot show A scan of the brain is not a photograph of thinking. MRI shows structure. fMRI measures the blood-oxygen signal (BOLD): active neurons use oxygen, blood flow rises a few seconds later, and the scanner sees the change. So it is indirect, slow (seconds) and blurry (millimetres), whereas EEG catches the brain's electrical rhythms in milliseconds but cannot say precisely where they come from. Neither shows a thought, and neither shows cause. For that you need a lesion, a drug, or brain stimulation. The classic warning is a dead Atlantic salmon. In 2009 Craig Bennett and colleagues put one in a scanner, "showed" it photographs of people, and found voxels (3D pixels) that lit up, because with tens of thousands of voxels, each tested at p < .05, some pass by chance. They used it to argue for corrections for multiple comparisons, and it won an Ig Nobel Prize in 2012. Two related traps: "reverse inference" (the amygdala was active, therefore the person felt fear: it is active in many states) and tiny samples. A 2022 study by Scott Marek and colleagues found that linking brain scans to behaviour in the general population needs thousands of people, not the dozens typical of older studies. Interactive · A scan of nothing p < .05p < .01p < .001 No correctionCorrected for all voxels Voxels in the scan Every voxel is pure noise. The grid shows 2,000 of the voxels; the readout counts them all. Bonferroni correction divides the threshold by the number of tests. With 50,000 voxels at p < .05 you expect 2,500 false "activations". Uncorrected, a scan of nothing finds a brain at work. Correcting the threshold for 50,000 tests keeps the chance of any false alarm near 5%. Modern studies correct, share data and replicate; but an old headline of the "this region lights up when…" kind deserves the same suspicion as any small study (see Research Methods in Psychology, 35/19). Sleep: a pressure that builds and a clock that ticks Sleep is not the brain switching off. Over a night it cycles about every 90 minutes through lighter and deeper stages and REM sleep, discovered in 1953 by Eugene Aserinsky and Nathaniel Kleitman, when the brain is highly active and most vivid dreams occur. A 2015 expert consensus recommended that adults regularly sleep at least seven hours; needs vary between people and across life, and no single "right" number fits everyone. The standard picture, proposed by Alexander Borbély in 1982, is the two-process model. Process S is a sleep pressure that rises the longer you are awake and drops during sleep (adenosine is a leading candidate for part of it, which is why caffeine, an adenosine blocker, delays sleepiness without removing the debt). Process C is the circadian clock, set mainly by a tiny region called the suprachiasmatic nucleus and by light, which makes you more alert at some hours than others. You feel sleepy when S is high and C is low. Jet lag and shift work are C and S out of step. Interactive · Pressure against the clock Sleep each night Bedtime Schematic model after Borbély, with rise time 18.2 h and fall time 4.2 h as in early human fits, scale 0 to 1. The circadian curve is a simple wave peaking at 18:00. Illustrative, not a prediction for any person. With 8 hours' sleep and bed at 23:00, pressure at waking falls to about 0.09 of the scale and ends the day near 0.62. Cut sleep to 5 hours and the morning level only gets down to about 0.22: you begin each day with leftover debt, and it does not clear itself overnight. Plasticity: the brain is rewired by use In 1949 Donald Hebb proposed that when one cell repeatedly helps to fire another, the connection between them strengthens. The slogan "cells that fire together wire together" is a later paraphrase. The effect, called long-term potentiation, was shown in rabbit hippocampus by Tim Bliss and Terje Lømo in 1973. It is the leading model of learning: experience changes connection strengths, not the number of cells. Interactive · A Hebbian cell Input A already drives the cell. Input B is too weak alone. Present them together and watch B's connection grow. Present A + BB aloneA aloneReset Rule: when the cell fires, each active input's weight w grows by 0.2 × (1 − w). The cell fires when the summed active weights reach 0.5. Cartoon of the Hebb rule, not a cell model. After three pairings B alone crosses the threshold. The adult brain stays plastic in far larger ways: in a 2000 study, London taxi drivers who had passed the demanding street test had a larger posterior hippocampus than controls, in proportion to years of driving. But claims that new neurons keep forming in adult humans are disputed. Two 2018 papers, one finding almost no new cells in adult hippocampus and one finding plenty, are still argued over. What is solid is that the brain changes with use throughout life, more in youth, and less than the popular word "rewiring" suggests. What a graduate just sees A neuron's message is a rate of identical spikes, not a loudness. Transmitters have many jobs. "The pleasure chemical" and "the chemical imbalance" are slogans. Nerve signals are fast and local; hormones are slow and body-wide. Functions run on networks, so a lesion shows what a region is needed for, not that it holds the skill alone. A brain scan is an indirect, slow, noisy measure and needs statistical care before it says anything. "We use 10% of the brain" and "left-brained people" are myths. Sleep pressure and the body clock are separate and add up to how awake you feel. Learning changes connection strengths, and use shapes the brain across life. Where it shows up next Sensation & Perception (35/21): how receptors turn light and sound into those spikes, and what the cortex does with them. Cognitive Psychology (35/22): memory and attention, and where the lesion cases above matter again. Clinical Psychology (35/27): mood, anxiety and psychosis, medication and therapy, where the transmitter story and its limits come back. Machine learning (34/32): artificial networks borrow the idea of weights strengthened by experience, loosely. In the world: caffeine and shift work, sleep clinics, brain-imaging headlines, anaesthesia, rehabilitation after stroke, and the question of how far any drug can explain a mood. If you want to go deeper 📘 Textbook. James Kalat, Biological Psychology (Cengage, 14th edition). Written for psychology majors; the common choice. 📘 Textbook. Mark Bear, Barry Connors and Michael Paradiso, Neuroscience: Exploring the Brain (Jones & Bartlett / Wolters Kluwer). More depth for neuroscience-leaning readers. 📘 Gentler. Oliver Sacks, The Man Who Mistook His Wife for a Hat (1985): case histories that make the lesion cases human. 🌐 Free course. MIT OpenCourseWare 9.01 Introduction to Neuroscience (Fall 2007). Also the introduction in 9.00SC Introduction to Psychology. The takeaway The mind runs on cells that signal by firing rate, chemicals that cross a gap, hormones that travel slowly, and connections that change with use. That is a powerful picture and a limited one: slogans about single chemicals or single brain regions, and strong claims from small brain scans, need the same scepticism as any psychological finding. Try this: this week, notice a mood shift (after poor sleep, a late coffee or a skipped meal) and think of the two layers: what the tissue is doing, and what you are feeling. Neither explains the other alone. Sources & Further Reading Sources. Action potential and leaky integrate-and-fire model: standard textbook (Kalat; Bear et al.); Hodgkin & Huxley (1952), Nobel 1963. Neuron count: Azevedo et al., Journal of Comparative Neurology (2009): about 86 billion neurons, about 80% in the cerebellum. Loewi (1921). Reward prediction error: Schultz, Dayan & Montague, Science (1997). Antidepressants: Cipriani et al., The Lancet (2018); Moncrieff et al., Molecular Psychiatry (2022), with published replies. Gage (1848), Broca (1861), Molaison (1953, Scoville and Milner). Left/right brain: Nielsen et al., PLoS ONE (2013). Dead salmon: Bennett et al. (2009); Ig Nobel 2012. Marek et al., Nature (2022). REM: Aserinsky & Kleitman (1953). Sleep duration: Watson et al., consensus statement, Sleep (2015). Two-process model: Borbély (1982); Daan, Beersma & Borbély (1984). Hebb (1949); Bliss & Lømo (1973). Maguire et al., PNAS (2000). Adult neurogenesis: Sorrells et al. and Boldrini et al., both 2018. Simulated numbers are toy models computed live. This is descriptive, not medical advice. If you are struggling, a doctor or local helpline (findahelpline.com) can help. 🌐 MIT OCW 9.01 Introduction to Neuroscience 🌐 MIT OCW 9.00SC Introduction to Psychology 📘 Kalat, Biological Psychology 📘 Sacks, The Man Who Mistook His Wife for a Hat Step 17. Bioinformatics & Genomics: Biology as Data --------------------------------------------------- Why: Genomic medicine and its data. A genome is a text of three billion letters that nobody can read in one piece. Biology now works by reading it in fragments, aligning the fragments, comparing them across species and people, and asking a computer what the pattern means. In April 2003 a headline said the human genome was finished. In March 2022 another said it was finished, again. Both were true. The first was a reference with gaps that held about 92% of the sequence; the second, 19 years later, filled in the rest. How can a text be "complete" twice? And how do you read three billion letters when your machine can only read a few hundred at a time? The answers are the core of bioinformatics: biology treated as information. A graduate sees a genome as a very long text with typos, repeats, copies and relatives, and sees biology questions as questions about strings: where does this read go, how similar are these two, which letters travel with a disease, and what shape does this sequence fold into? This piece walks through the five ideas that every course in the subject builds on. The course in one box Who takes itBiology, biochemistry, biomedical and computational-biology students, usually in year 3 or 4; computer-science and statistics students take it as an application. Some programmes teach it as a core course, others as a computing module inside genetics. What it is forTurning raw sequence data into biological statements: where a gene is, how two species are related, which variants are linked with a trait. Comes afterGenetics (35/46) and Programming Fundamentals (34/09). Leads toNothing later in this series, but it feeds modern Evolution (35/47), Microbiology (35/50), personalised medicine and any biology job that involves code. What this piece gives youYou will not be able to run a sequencing pipeline. You will see why reads must overlap, how an alignment is scored, how a tree comes out of a table of distances, why a hit in a genome-wide study is not a cause, and what AlphaFold does. Easier first reads: Entropy (13/12) for information ideas and LLMs (29/02) for pattern-finding by machine. A genome is a text read in scraps DNA is a four-letter text (A, C, G, T). A human haploid genome is about 3.1 billion letters, all copied in nearly every cell. No machine reads it end to end. Instead it is broken into pieces, every piece is read, and a computer stitches them back together by finding overlaps. This is shotgun sequencing, the method behind the genomes since the 1990s: in 1977 Frederick Sanger's group read the 5,386 letters of a small virus, phiX174; in 1995 the 1.8 million letters of the bacterium Haemophilus influenzae were the first free-living organism's genome; the Human Genome Project ran from 1990 to 2003, cost around $3 billion, and delivered a rough draft in 2000 and 2001 and a declared finish in April 2003. Since then the cost of sequencing has fallen far faster than the cost of computing did under Moore's law, which is why genomes of thousands of species and millions of people exist. Two numbers decide whether the assembly works. Coverage c is the average number of times each letter is read: reads × read length ÷ genome length. Reads land at random, so by the Lander–Waterman argument the fraction of the genome that nothing ever hit is e−c. uncovered fraction = e−cat 1× about 37% is missed; at 10× about 0.005%; at 30× essentially nothing The second number is the read length against the length of repeats. A read can only place itself if it contains something unique. The human genome is full of repeated stretches, some thousands of letters long and some, like the centromeres, millions long. A 150-letter read inside a 6,000-letter repeat has no way to know where it is. Long reads that reach out of the repeat can. Interactive · Shotgun sequencing: coverage and repeats Coverage c Read length (100 to 100,000 letters) Length of one repeat in the genome Genome 3,100 million letters. Reads = c × genome ÷ read length. Uncovered = e−c (idealised random reads, no errors). At 30× with 150-letter reads you need about 620 million reads, around 93 billion letters, and almost nothing is missed. But a 6,000-letter repeat is still unresolved. That is exactly why the 2003 "complete" genome kept holes, and why the 2022 genome needed reads tens of thousands of letters long. Alignment: finding the best way to line two texts up Once you have a sequence, the first thing you ask is: what does it resemble? Two similar sequences are rarely identical: letters are swapped, and some are inserted or deleted. An alignment lines two sequences up, with gaps, so that matching letters sit in the same column. Which alignment is best? Give points for matches, take points for mismatches, and take more for opening a gap, then find the arrangement with the highest total. In 1970 Saul Needleman and Christian Wunsch showed how to do that exactly, by filling in a table: S(i,j) = max( S(i−1,j−1) + score(aᵢ,bⱼ), S(i−1,j) + gap, S(i,j−1) + gap )best score for the first i letters of one and the first j of the other, built from three smaller answers This is dynamic programming: don't search every possible alignment (there are astronomically many), solve small prefixes once and reuse them. Smith and Waterman (1981) adapted it to find the best local match, and BLAST (1990) made a fast approximate version that searches databases of billions of letters in seconds. Try your own sequences below. Interactive · Needleman–Wunsch alignment Sequence 1 (letters A, C, G, T; up to 12) Sequence 2 GATTACA / GCATGCA One deletion Unrelated Match score Mismatch score Gap score Notice what the scores mean. Make gaps cheap and the program inserts gaps freely to force matches; make them dear and it prefers mismatches. There is no "true" alignment from the scores alone: the numbers encode a belief about how often evolution inserts, deletes and substitutes. Real tools use substitution matrices (BLOSUM, PAM) built from observed protein changes. Trees from distances: reading relatedness off a table Align many species and count the differences between each pair: you get a table of distances. Humans and chimpanzees differ at roughly 1 to 1.5% of aligned letters, humans and orangutans at about 3%, humans and macaques at about 6 to 7%. A tree is a way to draw such a table. The oldest recipe is UPGMA (Sokal and Michener, 1958): repeatedly join the two closest groups and average the distances. It is simple, and it quietly assumes a molecular clock: every lineage changes at the same rate. When rates differ, it goes wrong. In 1987 Naruya Saitou and Masatoshi Nei introduced neighbor-joining, which corrects each distance by how far each taxon is from all the others, so a fast-evolving "long branch" does not fool it. Interactive · The long-branch trap Four species A, B, C, D. The true tree pairs A with B and C with D. A and C evolve fast (long branches of length L); B and D are slow (length 1); the middle branch has length 1. (Illustrative lengths.) Length L of the fast branches With L=5 the table has AB=6, CD=6, AC=11, AD=7, BC=7, BD=3. The closest pair is B and D, at 3. UPGMA joins them, which is wrong: they are not each other's closest relatives at all, they only evolved slowly. Neighbor-joining recovers the true grouping. Below L=2 both agree. Modern methods (maximum likelihood, Bayesian inference) model rates explicitly, but the lesson stays: a method's assumptions are part of its answer. Genome-wide association: a million small bets If genomes differ by about one letter in a thousand between two people, which differences go with a trait? A genome-wide association study (GWAS) takes thousands to millions of people, records millions of variants (SNPs) and tests each one against the trait. The first big success was age-related macular degeneration in 2005. The statistics have a trap built in: test a million variants at p<0.05 and about 50,000 will "pass" by luck. The accepted threshold is therefore p<5×10−8 (roughly 0.05 divided by a million tests). A real association appears as a tall tower on a "Manhattan plot". Interactive · A simulated GWAS People in the study Variance each causal variant explains Naive threshold p < 0.05 Genome-wide p < 5×10⁻⁸ Simulation: 1,000 variants, 5 truly causal (marked ring), the rest null. Test statistic z ~ Normal(√(n·v/(1−v)), 1) for causal and Normal(0,1) otherwise. Fixed seed. Slide the sample size up and causal towers rise out of the noise; with the naive threshold the null variants light up too. Three cautions every course repeats. A hit is an association, often with a neighbouring variant that is inherited with it (linkage disequilibrium), not a proven cause. Most traits are polygenic: thousands of variants each with a tiny effect, so single hits explain little. And the studies have historically been dominated by people of European ancestry, so scores built from them predict less well elsewhere: a 2019 analysis of UK Biobank-based scores found them roughly two-fold less accurate in East Asian and several-fold (about 4.5 times in one summary) less accurate in African-ancestry samples. Biobanks in Japan, China, Mexico and Africa (BioBank Japan, China Kadoorie, the Mexico City Prospective Study, H3Africa) exist partly to fix this. What "heritability" does and does not mean is in Personality and Individual Differences (35/26). Also remember the human-variation fact: most genetic variation lies within populations, not between them (Lewontin, 1972, confirmed by later genomic work). This describes research and isn't medical advice. From sequence to shape: reading co-evolution, and what AlphaFold did A protein's function depends on its folded shape, and the shape is, in principle, written in its sequence. For decades predicting it was a grand challenge. One clue is old: if two positions in a protein touch in 3-D, a mutation at one is often compensated by a change at the other, so their letters change together across species. Align hundreds of relatives and look for columns that vary in step: that is co-evolution, and you can score it with mutual information. Interactive · Find the touching pair Sequences in the alignment Toy alignment: 10 columns, 4-letter alphabet. Columns 3 and 8 co-evolve (each value of one fixes the other); the rest vary independently. Fixed seed, first n rows used. With four or five sequences the pair you want is not even first: small samples make unrelated columns look linked. With about eight it leads, but rivals are still strong; with dozens or hundreds they fade and the true pair stands alone. That is why deep alignments mattered, and why DeepMind's AlphaFold 2 won the CASP14 competition in November 2020, scoring above 90 on the global-distance test for roughly two-thirds of targets, a level that experimentalists described as a stunning advance. It combines these alignment signals with a neural network trained on known structures (Jumper and colleagues, Nature, 2021). A public database of predictions launched in July 2021 with EMBL-EBI and grew to about 200 million structures in 2022. Demis Hassabis and John Jumper shared half of the 2024 Nobel Prize in Chemistry for it; David Baker took the other half for computational protein design. Limits matter. A predicted structure is a hypothesis with its own confidence score, not an experiment; the model does little for proteins that move, for disordered regions, or for how a mutation changes stability. Structural biologists still do the experiments, now with a far better starting guess. Privacy: your genome is also your family's Genetic data are unlike a password. You cannot change them, and they are shared with relatives who never agreed. In 2013 Gymrek and colleagues showed that surnames could be inferred from anonymous research genomes by matching Y-chromosome patterns against public genealogy databases. In 2018 investigators identified the Golden State Killer suspect by finding distant cousins in a public genealogy database; later that year Erlich and colleagues estimated that about 60% of searches for people of European descent in the US would already find a third cousin or closer in the consumer databases of the day, and that coverage would soon be near-universal. Genetic data are protected differently around the world: the EU's GDPR classes them as a special category needing stronger grounds; the US Genetic Information Nondiscrimination Act of 2008 covers health insurance and employment but not, for example, life insurance. When a testing company went into bankruptcy in 2025, many customers asked what happens to their data in a sale, which is a good question to ask of any provider. Indigenous communities have pressed for data sovereignty after cases such as the Havasupai blood samples, used for research they had not agreed to and settled in 2010. A genome is not just yours. Whoever holds your cousin's data holds a fuzzy copy of yours. What a graduate just sees A genome is a text; sequencing reads scraps and assembly depends on overlap, coverage (1 − e−c) and unique anchors. Repeats longer than the read are what stay unresolved, which is why "complete" took until 2022. Every alignment is a scoring choice. The numbers encode a belief about evolution. A tree is a model with assumptions. A clock assumption that fails puts the wrong species together. A million tests means a stricter threshold; a hit is a pointer, not a cause. A genome-wide result carries the ancestry of the people in it, so scores travel poorly. Sequence → shape works because shape is conserved and positions co-evolve. A prediction is a hypothesis. Genetic data identify relatives as well as you. Where it shows up next Genetics (35/46) and Evolution (35/47): variants, selection and phylogeny are now measured by exactly these tools. Microbiology (35/50): outbreaks and antibiotic-resistance genes are tracked by sequencing and tree-building, and metagenomics reads whole communities of microbes at once. Probability & Statistics (34/05): multiple testing, false discovery and confounding are the daily problems of a GWAS. In the world: rare-disease diagnosis, tumour sequencing, ancestry kits, forensic genealogy, crop and livestock breeding, and drug-target discovery. If you want to go deeper 📘 Textbook. Phillip Compeau and Pavel Pevzner, Bioinformatics Algorithms: An Active Learning Approach (Active Learning Publishers). Each chapter starts from a real biological puzzle and builds the algorithm. 📘 Textbook. Richard Durbin, Sean Eddy, Anders Krogh and Graeme Mitchison, Biological Sequence Analysis (Cambridge University Press, 1998). The classic on alignment and probabilistic models, still read. 📘 Gentler. Rosalind, a free site that teaches bioinformatics by short programming problems, from counting DNA letters up to assembly and alignment. 🌐 Free course. MIT OpenCourseWare 7.91J Foundations of Computational and Systems Biology (Spring 2014): sequence alignment, genomes and networks. The takeaway Biology is now also an information science: genomes are texts read in scraps, aligned by scoring, compared as distances, searched for associations and folded by learned patterns. Each step has a model with assumptions you should be able to name, and each step involves data about real people who are related to other real people. Try this: this week, take any word and a misspelling of it (for example "separate" and "seperate") and align them by hand: mark matches, a mismatch, and find where a gap would help. You have just run Needleman–Wunsch. Sources & Further Reading Sources. Human Genome Project: declared complete 14 April 2003, about 92% of the sequence, cost roughly $3 billion (Wikipedia summary of NHGRI and DOE records; NHGRI's sequencing-cost series, last updated 2023, graphs costs against a Moore's-law line). Telomere-to-telomere genome: Nurk et al., The complete sequence of a human genome, Science 376, 44–53 (31 March 2022; DOI 10.1126/science.abj6987), 3.055 billion letters and nearly 200 million new; the Y chromosome followed in 2023. Needleman & Wunsch 1970; Smith & Waterman 1981; Altschul et al. 1990 (BLAST); Sokal & Michener 1958 (UPGMA); Saitou & Nei 1987 (neighbor-joining); Lander & Waterman 1988 coverage argument. GWAS threshold 5×10⁻⁸ is standard; Martin et al., Nature Genetics 2019 (polygenic score accuracy by ancestry; secondary coverage gives slightly different African-ancestry ratios, so the figure here is approximate). AlphaFold: CASP14 (November 2020), Jumper et al., Nature 2021, database expansion to about 200 million structures (July 2022), 2024 Nobel Prize in Chemistry. Privacy: Gymrek et al., Science 2013; Erlich et al., Science 362, 690–694 (2018); GINA 2008; GDPR. The sequence tools and charts are teaching models computed in the page; the species distances are rounded and illustrative. This describes science and isn't medical or legal advice. 📘 Compeau & Pevzner, Bioinformatics Algorithms 📘 Durbin, Eddy, Krogh & Mitchison, Biological Sequence Analysis 🌐 Rosalind 🌐 MIT OCW 7.91J