iqbalr/Medical-MCQA

该数据集包含650个基于证据的医学多选题，采用JSON Lines格式存储。主要语言为英语，其中131条包含印尼语翻译。数据来源于PubMed和教科书式医学资源的检索增强生成流程，并经过验证。每个数据项包含医学主题标签、能力类别标签、问题题干、答案选项、正确答案、解释说明以及支持证据等核心字段。支持证据来自多个来源，包括PubMed(365条)、教科书(179条)和PubMed网页(113条)。数据集适用于医学多选题生成研究、基于证据的问答和双语医学教育实验等用途，但由于是机器生成和过滤的，不适用于直接临床决策。This repository contains 650 evidence-grounded medical multiple-choice questions in JSONL format. The dataset was prepared from a larger question bank generated with a retrieval-augmented generation pipeline that retrieves supporting evidence from PubMed and textbook-style medical sources, then verifies each item before logging it. Primary language is English with optional Indonesian translations for 131 rows. Each entry includes medical topic labels, competency categories, question stems, answer options, correct answers, explanations, and supporting evidence spans. Evidence sources include PubMed (365 entries), textbook (179 entries), and PubMed web (113 entries). The dataset is intended for research on medical MCQ generation, evidence-grounded QA benchmarking, and bilingual medical education experiments, but not for direct clinical use due to its machine-generated nature.