Can Stanford's AI predict future diseases from sleep?

Yes, Stanford's AI model, SleepFM, can predict over 130 different diseases, including dementia and cancer, from a single night of sleep data.

How accurate is the SleepFM AI in predicting diseases?

SleepFM achieved high accuracy, for instance, it predicted Parkinson's disease with 89% accuracy and dementia with 85% accuracy.

What is polysomnography and how does it relate to disease prediction?

Polysomnography is a sleep study that records various bodily functions; its data can be used by AI like SleepFM to predict future disease risks.

Home / Science / AI Predicts Diseases Years in Advance From Sleep

AI Predicts Diseases Years in Advance From Sleep

8 Jan

•

Summary

AI model predicts 130 diseases from one night of sleep.
SleepFM achieved high accuracy for Parkinson's, dementia, and cancer.
Sleep data may become a powerful tool for early disease detection.

AI Predicts Diseases Years in Advance From Sleep

Scientists have developed an advanced artificial intelligence program, SleepFM, capable of predicting an individual's risk for numerous diseases, including dementia, heart attack, stroke, and various cancers, potentially years before clinical diagnosis. The AI was trained on extensive sleep data from a large participant cohort, analyzing metrics such as brain waves, eye movements, and heart rhythm.

The model demonstrated significant accuracy in forecasting conditions like Parkinson's disease (89%), dementia (85%), and heart attacks (81%). Its predictive power extends to cancers, with 87% accuracy for breast cancer and 89% for prostate cancer, and even risk of death with 84% accuracy. This breakthrough suggests that polysomnography, a comprehensive sleep study, could evolve into a vital tool for proactive health monitoring.

Researchers are exploring enhancements, potentially integrating data from wearable devices. The findings, published in Nature Medicine, highlight sleep's profound connection to overall health and showcase foundation models' ability to interpret complex sleep data for early disease prediction. This innovative approach promises to transform how future health risks are identified and managed.