A multimodal sleep foundation model for disease prediction
A deep learning-based model, developed using the rich, multimodal data available from polysomnography-derived sleep recordings, performs well on common sleep analysis tasks and predicts future disease risk across a range of diseases.