Updated
Updated · Nature.com · Aug 3
Sleep AI Finds 5 Risk Groups, More Than Doubling Mortality Risk Over AHI
Updated
Updated · Nature.com · Aug 3

Sleep AI Finds 5 Risk Groups, More Than Doubling Mortality Risk Over AHI

3 articles · Updated · Nature.com · Aug 3

Summary

  • More than 10,000 clinical sleep recordings linked to electronic medical records let researchers train a foundation model that split patients into five risk groups with sharply different mortality, cardiovascular and neurological outcomes.
  • The highest-risk group, RG5, showed 2.38 times the mortality risk of the lowest group after adjustment, while conventional apnea-hypopnea index severity categories showed little or no predictive value.
  • The model learned from full-night polysomnography signals—not just breathing events—using multimodal data such as EEG, ECG, oxygen saturation and airflow, with sensitivity tests showing non-respiratory channels materially shaped risk assignment.
  • An external test in the Sleep Heart Health Study also separated higher- and lower-risk patients despite lower-resolution data, supporting the approach as a scalable path toward precision sleep medicine.

Insights

If AI can find deadly health risks in sleep data, will this make the traditional sleep apnea index completely obsolete?
Will transformer-based AI soon allow ordinary smartwatches to detect the same life-threatening sleep signals currently hidden in clinical lab data?
Could the hidden signals in your last sleep study predict your future risk for heart disease and cognitive decline?