Omni-Sleep: A Sleep Foundation Model via Hierarchical Contrastive Learning of CNS--ANS Dynamic

Short summary

Researchers introduce Omni-Sleep, a foundation model for sleep analysis that uses the brain and autonomic nervous system's physiological organization as a structural prior. Pre-trained on over 100,000 hours of polysomnography data across multiple centers, it outperforms existing models on sleep staging and multi-disease classification. Code is open-sourced.

• •Novel physiological hierarchy approach to sleep foundation models
• •Pre-trained on 100K+ hours of multi-center polysomnography data
• •Outperforms baselines on sleep staging and disease classification with improved cross-dataset generalization

Generated with AI, which can make mistakes.