Stanford’s new AI predicts over 100 diseases from just one night’s sleep. Stanford Medicine researchers have developed an artificial intelligence system, SleepFM, that can estimate a person’s risk of developing more than 100 diseases using data from just one night of polysomnography. Trained on about 585,000 hours of sleep recordings from roughly 65,000 patients, the multimodal “foundation model” learns patterns across brain activity, heart rhythms, breathing, muscle tone, and other physiological signals collected during overnight sleep studies. Once trained, SleepFM not only matched or surpassed current tools in conventional sleep tasks—such as staging sleep and assessing sleep apnea—but also identified long‑term health risks by linking sleep recordings with decades of electronic health records from the Stanford Sleep Medicine Center. Using this combined dataset, the model evaluated over 1,000 disease categories and found that 130 conditions could be predicted with meaningful accuracy based on sleep data alone. Its strongest performance, with concordance indices often above 0.8, was seen for Parkinson’s disease, dementia, hypertensive heart disease, heart attack, several cancers (including prostate and breast), pregnancy complications, mental health disorders, and even all‑cause mortality. The researchers discovered that no single signal was sufficient; instead, early warning signs emerged from mismatches across systems—for example, a brain that appears asleep while the heart looks awake. Ongoing work aims to improve accuracy, interpret how the AI reaches its conclusions, and potentially integrate wearable-device data, with the broader goal of turning routine sleep measurements into a powerful window on future health. References (APA style) Stanford Medicine. (2026, January 6). Stanford’s AI predicts disease risk from a single night of sleep. SciTechDaily. Thapa, R., Kjaer, M. R., Mignot, E., Zou, J., et al. (2026). A multimodal sleep foundation model for disease prediction. Nature Medicine.

Stanford’s new AI predicts over 100 diseases from just one night’s sleep. Stanford Medicine researchers have developed an artificial intelligence system, SleepFM, that can estimate a person’s risk of developing more than 100 diseases using data from just one night of polysomnography. Trained on about 585,000 hours of sleep recordings from roughly 65,000 patients, the multimodal “foundation model” learns patterns across brain activity, heart rhythms, breathing, muscle tone, and other physiological signals collected during overnight sleep studies. Once trained, SleepFM not only matched or surpassed current tools in conventional sleep tasks—such as staging sleep and assessing sleep apnea—but also identified long‑term health risks by linking sleep recordings with decades of electronic health records from the Stanford Sleep Medicine Center. Using this combined dataset, the model evaluated over 1,000 disease categories and found that 130 conditions could be predicted with meaningful accuracy based on sleep data alone. Its strongest performance, with concordance indices often above 0.8, was seen for Parkinson’s disease, dementia, hypertensive heart disease, heart attack, several cancers (including prostate and breast), pregnancy complications, mental health disorders, and even all‑cause mortality. The researchers discovered that no single signal was sufficient; instead, early warning signs emerged from mismatches across systems—for example, a brain that appears asleep while the heart looks awake. Ongoing work aims to improve accuracy, interpret how the AI reaches its conclusions, and potentially integrate wearable-device data, with the broader goal of turning routine sleep measurements into a powerful window on future health. References (APA style) Stanford Medicine. (2026, January 6). Stanford’s AI predicts disease risk from a single night of sleep. SciTechDaily. Thapa, R., Kjaer, M. R., Mignot, E., Zou, J., et al. (2026). A multimodal sleep foundation model for disease prediction. Nature Medicine.