Even personalized medicine will rely heavily on statistical relationships, just hopefully a little more nuanced than the crude racial generalizations you criticized above. How do these two genes usually interact in this context? How have other people with similar gene sequences (no two people are identical) responded to this drug or this vaccine in the past? Even the same body will respond differently at different times (as one of my relatives' ongoing struggle with her thyroid medicine shows).

There are some really cool advances. Aleksander Skarsgard's "body on a chip" models

https://steemit.com/science/@plotbot2015/tedx-guilford-college

are grown from the patient's own cells. Still, those are only predictive models in controlled lab conditions. They won't catch everything.

Science grows out of folklore, and is a refinement of folklore. The cutting edge is not magically sharp the way a lot of people imagine. It's fuzzy and fractal, as people struggle, one fact at a time, to determine what is true and what is not.

Most doctors do not get a lot of training in Bayesian statistics, and rely on simple heuristic rules based on their personal experience and some reading of the medical journals when they have time. Because there are statistical regularities in nature, these are often a good starting point. I agree that they are not a good stopping point, and that at least theoretically, the more information we have the better. The problem of which information is important is a hard one, though, and one that won't go away.