New AI System Analyzes Sleep for Early Disease Detection
SleepFM uses one night of sleep data to estimate risk for 130 diseases.
Polysomnography captures brain, heart, respiratory, eye, and muscle signals.
AI trained on 600,000 hours of polysomnography from 65,000 participants.
SleepFM uses five-second segments and contrastive learning for data harmonization.
C-index exceeds 0.8 for cancer, pregnancy complications, and mental disorders.
Researchers linked sleep records with 25-year health records for follow-up.
SleepFM requires interpretation and wearable integration for clinical use.
1 week ago