The Sleep Research Behind Our Condition Library

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).