Researchers at Palo Alto, Calif.,-based Stanford Medicine have developed an AI model capable of predicting an individual’s risk of developing more than 100 health conditions using sleep study data.
The model, called SleepFM, was trained on 600,000 hours of polysomnography data from 65,000 participants, according to a Jan. 6 news release from the health system and findings published the same day in Nature Medicine. It “excelled at predicting” the onset of Parkinson’s disease, dementia, hypertensive heart disease, heart attack, prostate cancer, breast cancer and death, researchers said.
Polysomnography is considered the gold standard in sleep diagnostics, capturing brain activity, heart signals, breathing patterns and other physiological metrics. SleepFM was trained using 585,000 hours of these recordings, split into five-second increments.
Stanford researchers paired the sleep recordings collected between 1999 and 2024 of 35,000 patients — ranging in age from 2 to 96 — with up to 25 years of follow-up data from electronic health records.
