YDXX/Risk-assessment-datasets

G-Health Risk Assessment Datasets

This repository contains the risk assessment training data used in G-Health, organized into four task types to support end-to-end health examination modeling.

Task Types Overview

1. Tabular Classification

The largest component, combining public and in-house datasets across a wide range of diseases and risk themes:

• —Diabetes (three complementary settings): pre-examination risk screening, classification based on examination results, and early-warning identification
• —Chronic disease & key-population risks: stroke risk, hypertension risk, obesity, maternal health risk
• —Screening & long-term management: nasopharyngeal cancer, cervical cancer, chronic kidney disease, thyroid cancer, sleep disorders, heart failure, hepatitis

2. Regression

Continuous prediction tasks:

• —Cardiac risk assessment: maps examination indicators to continuous risk scores or ordinal risk levels
• —Calorie-related prediction: estimates energy intake and expenditure for personalized health management

3. Image Classification

Three visual modalities central to health examinations:

• —Ophthalmic imaging: multi-label, multi-class detection (DR, DN, MH, ODC, TSLN, normal findings)
• —Dermatology images
• —Chest CT

4. Object Detection

• —Breast cancer detection: lesion localization and highlighting for risk alerts and referral suggestions

Data Sources

Tabular Classification

Regression

Image Classification

Object Detection

Dataset Structure

Risk-assessment-datasets/
├── csvfile/          # Tabular & regression data (CSV)
│   ├── diabetes_dataset.csv
│   ├── stroke_risk_dataset.csv
│   ├── hypertension_risk_dataset.csv
│   ├── heart_risk_dataset.csv
│   ├── calories_dataset.csv
│   └── ...
└── Img/              # Image data
    ├── Breast_cancer/    # Object detection
    ├── eye_disease/      # Ophthalmic
    ├── Skin_Cancer/      # Dermatology
    └── chest_cancer/     # Chest CT

Citation

@article{lin2026clinically,
  title={Clinically grounded multi-agent artificial intelligence for preventive health management},
  author={Lin, Hao and Zhang, Yang and Ye, Dongxin and He, Sicheng and Du, Zhaowu and Yu, Yang and Yu, Xiao and Ren, Liping and Dong, Nanqing and Hu, Fang and others},
  year={2026}
}

Citation

@article{lin2026clinically,
  title={Clinically grounded multi-agent artificial intelligence for preventive health management},
  author={Lin, Hao and Zhang, Yang and Ye, Dongxin and He, Sicheng and Du, Zhaowu and Yu, Yang and Yu, Xiao and Ren, Liping and Dong, Nanqing and Hu, Fang and others},
  year={2026}
}