Sleep Domain’s condition library draws on findings from SleepFM, a Stanford Medicine AI model that analyzed overnight sleep-study (polysomnography) recordings from tens of thousands of patients alongside their long-term health records. The researchers found that signals recorded during a single night’s sleep were associated with the future onset of a wide range of health conditions, evaluated years in advance.
What this does not mean: SleepFM is a research model, not a diagnostic tool, and it has not been approved for clinical use. The associations it identified describe population-level statistical patterns, not individual predictions. Nothing on this site is medical advice — talk to a qualified clinician about your own risk factors and symptoms.
Citation
Thapa, R., Kjaer, M.R. et al. “A multimodal sleep foundation model for disease prediction.” Nature Medicine (2026). https://rdcu.be/yUKJRS43SUSa (DOI: 10.1038/s41591-025–04133‑4)
See also the Stanford Medicine news summary: “New AI model predicts disease risk while you sleep” (Jan. 6, 2026).