Can a Single Night of Sleep Predict Future Diseases?
- Seoyoung Kang
- Feb 24
- 1 min read
Feb 24, 2026
Seoyoung Kang
Researchers at Stanford University in the United States have developed an artificial intelligence (AI) model called SleepFM that can predict future disease risks using data from just one night of sleep testing. The findings were published in the international medical journal Nature Medicine in January 2026.
SleepFM analyzes brain waves, heart rate, breathing patterns, and muscle movements recorded during sleep to estimate the risk of about 130 diseases, including Parkinson’s disease, dementia, heart disease, and breast cancer. One particularly notable finding is that the model is able to detect warning signs of disease several years before symptoms begin to appear.
The research team trained the AI using sleep study data from approximately 65,000 individuals, along with 25 years of electronic medical records. This helped the system identify statistical links between sleep patterns and later disease diagnoses.
However, experts state several limitations. The AI cannot explain the causes of diseases and is mainly based on data from patients who visited sleep clinics for sleep-related issues. As a result, its accuracy for the general population may be limited.
This study suggests that sleep data can be used not only to assess fatigue but also to predict long-term health risks. However, because the data mainly comes from patients at sleep clinics, further research involving the general population will be necessary.
