🩺Medical benchmarks measure if LLMs get the correct final diagnosis. True clinical reasoning requires sequential belief updating: does the model revise its beliefs appropriately as new evidence appears?

New preprint: arxiv.org/abs/2505.22919 We introduce ER-Reason, a dataset of 25,174 de-identified clinical notes from 3,437 patients that supports evaluation across all stages of the emergency department (ED) workflow: triage intake, treatment selection, disposition planning, and final diagnosis.