Scientists created SleepFM, an artificial intelligence system that analyzes polysomnography data from 585,000 hours of sleep recordings across 65,000 participants to predict disease risk. The model achieved C-indices exceeding 0.75 for 130 medical conditions, including mortality (0.84), dementia (0.85), and heart attack (0.81) from just one night's sleep data. This breakthrough represents a paradigm shift in sleep medicine by demonstrating that complex physiological patterns during sleep contain predictive signatures for future health outcomes years before clinical symptoms emerge. The model's ability to extract disease-relevant information from the intricate interplay of brain waves, breathing patterns, and cardiac rhythms during sleep suggests sleep architecture serves as a biomarker-rich window into systemic health. While promising for early disease detection and prevention strategies, the approach requires validation across diverse populations and healthcare settings before clinical deployment. The technology could eventually enable routine sleep studies to function as comprehensive health screenings, transforming how we approach preventive medicine and early intervention.