Sleep is essential to health—so essential that complete sleep deprivation will be fatal within a relatively short period of time. Suboptimal or poor sleep can also contribute to (or result from) many different diseases. But sleep itself is a complex process that involves every organ system in the body.
That complexity can be captured with polysomnography, a comprehensive overnight study that records hours of interconnected signals—brain activity, eye movements, respiration, heart rhythm (EKG), and more. The obvious question is: could this massive dataset contain subtle clues or patterns that foreshadow future illness?
To find out, Stanford researchers trained an AI model on polysomnography data from 65,000 people and linked those recordings to the same individuals’ health records. They found that subtle patterns in a single night of sleep data could predict the future development of 130 diseases, including dementia, heart failure, and kidney disease. Looking ahead, this tool could shift care earlier—helping people reduce risk by making lifestyle changes or starting treatment early, and enabling researchers to develop interventions that stop disease before it takes hold. Because as powerful as treatment can be, prevention is even better.
