It analyses brain waves, heart activity, breathing, and movement recorded during polysomnography. SleepFM was trained on nearly 600,000 hours of data. Researchers say it predicts risks years before symptoms appear, but only identifies correlations, not causes.

Credits: envato elements; Author: Rawpixel;

Analysis shows that heart signals are most predictive of cardiovascular disease, while brain signals are more predictive of neurological disorders. Experts stress AI supports, but does not replace, medical judgment.