cff-version: 1.2.0
message: "If you use this dataset, please cite it as below."
title: "BedsideBench: A Clinical Question-Answering Benchmark"
abstract: >-
  A 500-case clinical question-answering benchmark covering medication safety,
  guideline and trial evidence, false-premise queries, quantitative reasoning,
  health equity, and free-text management plans. The release includes public
  grading rubrics and a reproducible LLM-judge protocol.
type: dataset
version: "1.0.0"
date-released: "2026-09-14"
authors:
  - name: "Doximity"
license: CC-BY-NC-SA-4.0
keywords:
  - clinical NLP
  - medical question answering
  - LLM evaluation
  - benchmark
  - patient safety