🛌 AI predicts 130 diseases from a single night of sleep
A new study in Nature introduces SleepFM, a foundation model trained on 585,000 hours of polysomnography (PSG) from 65,000 participants. Unlike previous AI that only stages sleep or detects apnea, SleepFM creates a general sleep embedding from a single night of data.
How it works
• PSG captures brain, heart, muscle, and breathing signals during sleep.
• SleepFM resamples signals to 128 Hz, splits them into 5-second tokens, uses 1D convolutions, and pools channels via attention before a transformer summarizes the full night.
• Leave-one-
A new study in Nature introduces SleepFM, a foundation model trained on 585,000 hours of polysomnography (PSG) from 65,000 participants. Unlike previous AI that only stages sleep or detects apnea, SleepFM creates a general sleep embedding from a single night of data.
How it works
• PSG captures brain, heart, muscle, and breathing signals during sleep.
• SleepFM resamples signals to 128 Hz, splits them into 5-second tokens, uses 1D convolutions, and pools channels via attention before a transformer summarizes the full night.
• Leave-one-
