临床指南与专家共识数据集

本数据集面向医学垂类大模型研发团队、临床决策支持系统开发商、循证医学知识库建设方，以及需要分级推荐强度的权威临床指南语料支撑智慧医疗产品的企业，旨在解决临床指南与共识文档分散、推荐强度缺乏统一标注、跨学会跨语种指南难以系统化归集等问题，这些问题直接导致医学AI在循证诊疗推理中出现证据等级混乱。数据集按推荐强度系统分级为临床指南、专家共识、诊疗建议三大类别，涵盖呼吸科、感染科、风湿免疫科、肿瘤科、皮肤科、血液科、神经科、影像科、检验科等数十个专科领域。每篇均标注发布机构、指南等级、发表年份、学科分类及原文链接，语种覆盖中、英、德、日、意等多语种。与传统零散收集的指南文档不同，本数据集已完成跨学会、跨语种的系统化归集与推荐强度统一标注，将非结构化的PDF指南转化为可直接用于模型训练的分级知识单元。

This dataset is intended for teams developing medical domain-specific large language models, developers of clinical decision support systems, constructors of evidence-based medicine knowledge bases, and enterprises that need authoritative clinical guideline corpora with graded recommendation strength to support their smart healthcare product development. It aims to address issues including scattered distribution of clinical guidelines and consensus documents, lack of unified annotation for recommendation strength, and challenges in systematic aggregation of cross-society and cross-language guidelines, which directly cause confusion in evidence hierarchy during evidence-based diagnostic and therapeutic reasoning for medical AI. The dataset is systematically categorized by recommendation strength into three major categories: clinical guidelines, expert consensuses, and diagnostic and therapeutic recommendations, covering dozens of specialized fields such as Respiratory Medicine, Infectious Diseases, Rheumatology and Immunology, Oncology, Dermatology, Hematology, Neurology, Radiology, and Clinical Laboratory Medicine. Each entry is annotated with publishing institution, guideline level, publication year, subject classification, and original document link, and supports multiple languages including Chinese, English, German, Japanese, and Italian. Unlike traditionally scattered collections of guideline documents, this dataset has completed systematic cross-society and cross-language aggregation and unified annotation of recommendation strength, converting unstructured PDF guidelines into graded knowledge units that can be directly utilized for model training.

搜集汇总

背景与挑战

背景概述

该数据集系统化归集了跨学会、跨语种的临床指南与专家共识，按推荐强度分为临床指南、专家共识和诊疗建议三大类，覆盖呼吸科、肿瘤科等数十个专科领域。每篇文档均标注了发布机构、指南等级、发表年份及学科分类等信息，并将非结构化PDF转化为可直接用于AI模型训练的分级知识单元，旨在支持医学大模型和临床决策系统的循证医学应用。