'Medical malpractice' is different to 'malpractice' in general. The first is deviations from standard care the second is a dereliction of duty. I think it is every doctors duty to get good at statistics. Clearly the Bayesian statistics classes have not been sufficient.

If you consider the role of doctors to be finite state automata that execute instructions handed to them from standards bodies then we wouldn't need anything nearly as sophisticated as a LLM to replace them, a collection of decision trees would be sufficient.

Your input on the topic has not dissuaded me from my belief that doctors in general are bad at stats and therefore bad at their jobs.