Internist at Lahey practicing “Less is More” Medicine | Critical Appraisal EBM | Dog father | ADHD advocate | Aspiring Polymath | Views and Opinions my own |
@adamcifu For me, it's a point in the visit where my questions are opening a Pandora's box of more questions. Unfortunately, there is no going back from there.
@AdamRodmanMD Interesting! Is there a link or the title of the pre-print? o1 more often might be relevant to coding or other domain specific tasks vs probably clinical reasoning. For me, 4o does the work.
@AdamRodmanMD I often use the Socratic method prompt, to question my own assumptions, and reasoning. I wonder if they can get trained similarly. Created a custom GPT based on it, that works fantastic!
@AdamRodmanMD I think of Clinicians as natural Bayesians that get trained over time, with our own systematic biases built in.LLMs obviously have better Bayesian predictions on paper but it makes sense how the underlying neural network is akin to our brain’s network. A bit reductive, obviously
@NTFabiano Yeah. I think of it as part of a spectrum. So many of our chronic physical diseases are due to underlying undiagnosed mental health issues that are being addressed in maladaptive ways. I hate how medicine is siloed in its ways, that we forget it’s all connected.
@NTFabiano A lot of times depression has its underlying causes, and can have often undiagnosed ADHD, or other conditions. I often see a spectrum of conditions, with often significant overlap. I think, there is a huge potential for novel medications, but also recognizing the entire picture!