Oddbean new post about | logout
 I'm not saying anything about the validity of AI in medical or even your interest in the subject, but after listening to the other side, all I could think was, "but, why?".  Supposedly, standard treatment paths were supposed to be the "AI" of yesteryear, the culmination of centuries of medical practice and knowledge, coalesced into the physical form of textbooks, manuals and education, yet somehow, diseases go misdiagnosed, improper and outright wrong surgeries and procedures are performed, talk of cure is forbidden and censured while diet and nutrition are sidelined as not being causal...
Perhaps AI could distill centuries of medical wisdom into the hands of you and me... Would we act responsibly with these new powers?

#NotControversial

#JoelDWallach

#DeadDoctorsDontLie      #LetsPlayDoctor

#WhereThereIsNoDoctor

#SelfCare      #SickCare

#FireYourDoctor

#GrowNostr 
 So what I understand about AI in med, for a specific example, take radiology.

AI has been able to detect breast cancer in women earlier than the human eye can.

That’s just one example. 
 Excellent!  Finding a signal in what would be otherwise perceived as noise by diagnosticians... Yes, I agree, this IS a proper application for AI. 
 Yes AI in radiology use-case are the most developed and show the direction of travel.  Same goes for digital pathology.  Of course there are error messages with false positives etc, but that needs to be countered against the actual real-world accuracy of clinicians in the field. 
 There is an argument that many clinical presentations are compromised by the need for a formal diagnosis before those standards kick in.  Let's take the spectrum of metabolic disorders that culminate with a final diagnosis after years of growing sub-clinical deterioration. I am also 100% with you that health systems have been poor implementing standardized best practice care.