HEALIX
收藏资源简介:
HEALIX是由美国国家医学图书馆联合多机构开发的全球首个公开临床笔记健康素养标注数据集,包含来自MIMIC-III数据库的589条跨9种类型的临床笔记,通过社工笔记抽样、关键词过滤及大模型主动学习三重策略构建。数据集采用三级标注体系(低/正常/高健康素养),旨在解决传统健康素养筛查工具在电子健康记录中难以结构化记录的痛点,为自然语言处理模型开发提供基准资源,可应用于患者教育干预、住院再入院率预测等医疗决策支持场景。
HEALIX is the world's first publicly available clinical note health literacy annotation dataset developed by the U.S. National Library of Medicine in collaboration with multiple institutions. It contains 589 clinical notes spanning 9 categories from the MIMIC-III database, and was constructed using three strategies: social worker note sampling, keyword filtering, and large model active learning. The dataset adopts a three-level annotation system (low/normal/high health literacy), aiming to address the pain point that traditional health literacy screening tools struggle with structured documentation in electronic health records. It provides benchmark resources for natural language processing model development, and can be applied to medical decision support scenarios such as patient education intervention and hospital readmission rate prediction.
HEALIX数据集概述
数据集名称
HEALIX
核心用途
用于从临床记录中识别患者健康素养信息。
访问方式
需要通过Physionet进行凭证访问。
数据文件
- 主要数据文件:
healix.json
加载方法
使用提供的脚本加载
执行命令:python loading_HEALIX.py
在自定义脚本中加载
python from loading_HEALIX import load_healix data = load_healix("HEALIX.json")
运行环境要求
- Python 3.x




