Two patients. Same diagnosis. Same drug. One gets better. The other doesn't respond at all.
It sounds like a rare case. It's not. It's actually pretty common.
Medicine, for the most part, has always worked with averages. What helps the majority? What works on paper. But bodies don't work on paper — yours processes medication differently than mine, and differently again from the person sitting next to you. Genes play a role. So does gut health, stress, sleep, a dozen other things we're only starting to measure properly.
AI is starting to change this. Slowly, but it's happening.
Machine learning can now go through thousands of biological variables and give a prediction — how will this specific person respond to this specific drug? Not after the side effects hit. Not after the second failed prescription. Before any of that.
Tempus and Recursion Pharmaceuticals are already doing this in cancer care, where there's no time to guess.
Denmark sits in an interesting position here. The Danish National Patient Registry has data on over 8 million people going back decades — one of the richest health datasets anywhere. Researchers at DTU and the University of Copenhagen are working on what becomes possible when AI meets that kind of data. Something that actually fits you, not just most people.
We're not there yet. But honestly, we're closer than I expected.
Would you trust an AI to choose your medication — or does that feel like a step too far?
