MENTAT
收藏资源简介:
MENTAT数据集是由美国的精神科医生专家创建和注释的,包含五个关键决策领域的实际问题:治疗、诊断、文档记录、监测和分诊。该数据集旨在捕捉精神科医生在日常护理中遇到的临床推理细微差异和日常模糊性,反映了现有数据集中缺失的护理交付的内在复杂性。数据集包含203个基础问题,每个问题有五个答案选项,且去除了与决策无关的病人人口统计信息,适用于男性、女性或非二元编码的病人。对于涉及模糊性和多个有效答案选项的问题类别,创建了一个带有专家注释不确定性的偏好数据集。
The MENTAT dataset was developed and annotated by American psychiatric experts, encompassing real-world clinical questions across five core decision-making domains: treatment, diagnosis, documentation, monitoring, and triage. This dataset is designed to capture the nuanced differences in clinical reasoning and everyday ambiguities faced by psychiatrists in routine clinical care, reflecting the inherent complexities of care delivery that are missing from existing datasets. The dataset contains 203 foundational questions, each with five answer options, and eliminates patient demographic information irrelevant to decision-making, making it applicable to male, female, or non-binary coded patients. For question categories involving ambiguity and multiple valid answer options, a preference dataset annotated with expert-derived uncertainty was created.

- 1Moving Beyond Medical Exam Questions: A Clinician-Annotated Dataset of Real-World Tasks and Ambiguity in Mental Healthcare斯坦福大学, 科罗拉多大学, 西北大学, 威斯康星大学, 耶鲁大学, 芝加哥大学, 俄亥俄州立大学 · 2025年



