Bottom-up medicine
Draft, still being revised
My archive holds my old homepage, which I had not looked at in twenty years. On the science page there is a box about medicine. After my civil service in a hospital I wrote there: “i think, medicine is in many cases wrong and a lot has to be done to correct this: building up new theories, more and different research in physiology and cellbiology and of course: development of new diagnostic instruments.”
I no longer know what exactly I was thinking back then. Today I see it like this: medicine is one of the last sciences that is not built from the bottom up, the way mathematics or physics are. How anaesthesia works at the molecular level is still not fully understood today, and the same goes for many other drugs. They were tested, experience was collected, statistics were done. We are fairly sure that something works, but we do not fully understand why.
That is changing now. AlphaFold, a tool that predicts the structure of proteins, is only a few years old. AlphaFold does not design drugs itself, but once you know the shape of a protein, you can build a compound that fits it. That is what people are now starting to do with AI: design drugs deliberately instead of searching for them. Incidentally, the mRNA vaccines against Covid are another example: they were built on purpose, after the virus was understood in detail. There will be such vaccines against cancer too, tailored to individual people. A personalised mRNA vaccine against melanoma met its goal in a phase 3 trial in 2026.