4 ms·

> if ChatGPT has a 50% chance of giving the wrong diagnosis It depends on the average chance of the doctor giving the wrong diagnosies (I honestly have no idea

• OJFord 3y ago
  My experience is second-hand, but they seem to be pretty split between seeing exciting rare diagnoses everywhere (students, juniors primarily, or one's own particular research interest) and biased against them, it's ~never that, it's the same boring thing we saw ten of yesterday (jaded specialists primarily).
• EA-3167 3y ago
  There are different figures, partly as a result of how this question is studied. The Mayo Clinic did a famous study on people who sought a second opinion, and they found that something like 12% of people seeking that second opinion were correctly diagnosed. Obviously this isn't a fair sample since this is not only people seeking a second opinion, but one from a prestigious source. This narrative review discusses the issue from a global perspective, concluding that most people will experience at least one misdiagnosis. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5502242/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5502242/ Article on Mayo study, with link to study: https://newsnetwork.mayoclinic.org/discussion/mayo-clinic-researchers-demonstrate-value-of-second-opinions/ https://newsnetwork.mayoclinic.org/discussion/mayo-clinic-re...
  • einpoklum 3y ago
    Isn't seeking a second opinion something you usually do when the first opinion sounds fishy, doesn't seem to match the evidence, or when the doctor giving doesn't seem to be doing his job seriously (e.g. dismissive, preoccupied etc)? Or in cases where it is known to be a difficult situation which makes you think multiple diagnoses are needed?