Career path: AI drug discovery or medical AI?
1 antoperezmed 1 9/5/2025, 12:49:24 PM
I am a med student learning to code and planning to get into ML research applied to medicine, but not sure which of those makes more sense to get into in the long term. Drug discovery seems more complex science wise (the field is full of PhDs), whereas medical AI (medical imaging, EHRs, etc) seems to have its bottleneck in the regulations, politics-economics and lack of trust from doctors. Anyone here working on either of them that can share their thoughts?
Don't sell pharma short when it comes to regulations, politics-economics and lack of trust from doctors :\
Are you thinking of switching to Pharmacy and following the Medicinal Chemistry track?
Regardless, I like the idea of advancing in natural science in a way that compounds over the years whether still in school or not.
Also it would probably be best if you can make sure to be able to do fine without AI to begin with, keeping in mind the whole time how you would use it to leverage the things where you already know it could be helpful. Along with things that maybe nobody else has thought of. Keeping up with the limitations & abilities of AI as well as your chosen field, relative to the resources you have and what you can make of them.
For instance when you become a physician you really should feel 100% confident that you could go forward indefinitely treating patients without any AI at all, or as a drug developer create or isolate at least one novel compound on your own the old fashioned way which would sit in the market at a very profitable place. That's pretty good validation right there that you would be able to make more of AI, than you would without that hands-on ability. Nobody can expect your first viable compound to be effective, safe, or approvable, much less a blockbuster, that's something completely different.
From that point, going whole-hog into AI and/or research could take you a lot further, especially if there is not much more front-line hands-on effort with the patients or chemicals going forward.
Keep in mind that with pharma more than ever it's like pop music and depends on highly promotable "viral hits" more so than lots of other health-care areas, so most things never make it to the big time even when the talent is right up there.
I would say there is already plenty of domain expertise that can be leveraged by AI, but if you build then leverage your own personal domain expertise, or add it to the mix it should add up to quite a bit more.