A revolutionary AI model could predict about 130 serious diseases using data from just one night of sleep. Developed by scientists at the Stanford University, this breakthrough marks a milestone in predictive medicine.
The birth of SleepFM: AI model that analyzes sleep
Stanford researchers introduced SleepFM, the first AI model trained on nearly 600,000 hours of sleep data from 65,000 individuals. It uses polysomnography, the gold standard for measuring brain waves, heart rhythms, muscle activity, and breathing during sleep.
Accuracy in high‑risk predictions
This AI model identifies risks of dementia, myocardial infarction, heart failure, chronic kidney disease, strokes, and atrial fibrillation. It generates latent representations that capture the physiological structure of sleep, overcoming barriers in massive complex‑data analysis.
Historical context of AI use in sleep
Before SleepFM, sleep remained under‑studied in AI despite its link to chronic health. Previous studies, such as those by Mass General Brigham in 2021, used AI to detect type‑2 diabetes with wearables, but none scaled to 130 pathologies with polysomnography. James Zou, associate professor of biomedical data at Stanford, notes: “Sleep is relatively understudied from an AI perspective”.
Implications for global health
The publication in Nature Medicine validates SleepFM as a pioneer in large‑scale analysis. It could transform preventive check‑ups, providing early alerts about high mortality. WHO statistics indicate that cardiovascular diseases cause 17.9 million deaths annually; such a AI model could reduce them through early detection.
Limitations and future
While promising, it requires validation in diverse populations. Experts like Zou emphasize its potential to integrate AI into daily clinical routines.
- Training: 65,000 participants.
- Data: 600,000 hours of PSG.
- Diseases: 130, focused on lethal.
This AI model not only predicts, but also redefines sleep monitoring. In the Dominican Republic, where sleep disorders affect 30% of adults according to local surveys, its impact could be transformative. SleepFM positions the AI model as a key ally in future public health.
