Patterns, a Cell Press journal (@cp-patterns.bsky.social)

AISleep: Automated and interpretable sleep staging from single-channel EEG data Sleep staging is vital for understanding sleep physiology and diagnosing sleep disorders, yet manual scoring is time-consuming and unscalable. The authors introduce AISleep, an unsupervised sleep staging algorithm based on feature-weighted kernel density estimation, using only a single EEG channel. AISleep outperforms SOTA unsupervised methods, generalizes better than supervised models, and uncovers age-related EEG feature decline impacting staging accuracy. Its lightweight design enables integration into portable devices, providing a practical, scalable solution for accurate and accessible home-based sleep monitoring. http://dlvr.it/TN5N9d