Stanford AI Model Uses Sleep Data to Predict Disease Risk

Poor Sleep, Long-Term Clues

A restless night may leave you groggy the next day, but new research suggests it could also reveal health risks years in advance. Scientists at Stanford Medicine and their collaborators have developed an artificial intelligence model capable of predicting a person’s likelihood of developing more than 100 medical conditions based on a single night’s sleep.

Introducing SleepFM

The model, called SleepFM, was trained on nearly 600,000 hours of sleep recordings from 65,000 participants. The data came from polysomnography, the gold standard in sleep studies, which uses sensors to track brain activity, heart rhythms, breathing, eye movements, and muscle activity overnight. Researchers realized this wealth of physiological information was largely untapped in traditional sleep medicine.

Unlocking Hidden Data

“We record an amazing number of signals when we study sleep,” said Emmanuel Mignot, MD, PhD, Craig Reynolds Professor in Sleep Medicine and co-senior author of the study, published January 6 in Nature Medicine. “It’s a kind of general physiology that we study for eight hours in a subject who’s completely captive. It’s very data rich.”

Until now, only a fraction of this data was analyzed. Advances in AI have made it possible to process the full scope of signals, uncovering patterns linked to long-term health outcomes.

Sleep as a Frontier for AI

“From an AI perspective, sleep is relatively understudied,” noted James Zou, PhD, associate professor of biomedical data science and co-senior author. “There’s a lot of other AI work that’s looking at pathology or cardiology, but relatively little looking at sleep, despite sleep being such an important part of life.”

Looking Ahead

The study marks the first large-scale use of AI to analyze sleep data, opening new possibilities for early detection of disease and personalized health monitoring.