DischargeBench:以患者人设模拟评估 LLM 出院教育能力
原标题:From Discharge Notes to Patient Understanding: Persona-Grounded, Open-Ended Simulation of LLMs as Discharge Educators
AI 摘要
马萨诸塞大学洛厄尔分校与阿默斯特分校的研究者提出 DischargeBench,一个以患者人设为基础、开放式多轮对话的模拟框架,用于评估 LLM 作为出院教育者的能力。他们构建了包含 477 个病例、覆盖 24 个 ICD 章节的 MIMIC-IV-Ext-DischargeBench 数据集,并引入教育监督智能体来维持虚拟患者的真实感。评估从对话质量、主题清单、理解程度和事实一致性四个维度打分,结果显示总体分数掩盖了不同 ICD 章节和患者人设之间的临床相关差异,困难人设会暴露覆盖失败和理解缺口。
正文节选
From Discharge Notes to Patient Understanding: Persona-Grounded, Open-Ended Simulation of LLMs as Discharge Educators Abstract Hospital discharge education is an interactive teaching task: a clinician adapts a discharge plan to a patient’s literacy, recall, and personality. Existing LLM evaluations target static or artifact-generation tasks and do not measure patient understanding under open-ended dialogue. We introduce DischargeBench, a persona-grounded simulation in which a candidate LLM educ