AI健康指数与疾病预测训练数据集——超1亿影像记录+2亿结构化指标的多模态机器学习库

本数据集专为AI大模型训练、疾病早筛算法开发及风险评估模型优化设计，是国内规模领先的多层次健康体检训练数据集。数据总量包含超1亿人次影像学检查记录（涵盖低剂量螺旋CT、超声、心电等）与超2亿人次结构化检验指标数据，部分关键字段附带专家标注的重要异常结果标签（标签检出率约10%）。

This dataset is purpose-built for LLM training, disease early screening algorithm development and risk assessment model optimization, and is a leading multi-level health examination training dataset in China in terms of scale. The total dataset includes over 100 million person-times of imaging examination records (covering low-dose spiral CT, ultrasound, electrocardiogram (ECG), etc.) and over 200 million person-times of structured laboratory test data. Some key fields are accompanied by expert-annotated labels for important abnormal findings, with a label detection rate of approximately 10%.

搜集汇总

背景与挑战

背景概述

该数据集是一个专为AI大模型训练和疾病预测设计的多模态机器学习库，包含超1亿影像记录和超2亿结构化指标，覆盖从健康到多病共患的全谱系人群，并附带专家标注和时间序列数据。数据集已完成匿名化脱敏，支持定制化交付，适用于健康风险评估和算法开发。