Foundation Model Uses Rich Sleep Data to Predict Disease Risk and Survival

EricTopol · x · 2026-08-03

A recent study published in Nature Communications introduces a foundation model for sleep-based risk stratification. By learning rich representations from over 10,000 clinical sleep recordings linked to electronic medical records, the model extracts deep physiological insights.

The research reveals that sleep physiology contains a latent risk structure invisible to conventional metrics like the apnea–hypopnea index. Using this structure, the model successfully identified five distinct patient risk groups and effectively predicted future disease risk and clinical survival rates. This marks the second recent foundation AI model to demonstrate the immense potential of rich sleep data in predicting medical outcomes.

Original post →

More from Research

Research channel →