ActiTect: a generalizable machine learning pipeline for REM sleep behavior disorder screening through standardized actigraphy - PubMed
4 hours ago
- #actigraphy
- #REM sleep behavior disorder
- #machine learning
- ActiTect is an automated, open-source machine learning pipeline for screening REM sleep behavior disorder (RBD) using actigraphy.
- It includes robust preprocessing and sleep-wake detection to harmonize multi-device data and extract interpretable motion features.
- Model development achieved an AUROC of 0.95 in a cohort of 78 individuals, with strong generalization to local and external test sets.
- Leave-one-dataset-out cross-validation showed consistent performance (AUROC range 0.84-0.89), and feature reproducibility supports a pre-trained model for broader deployment.
- The tool is designed to advance generalizable wearable-based RBD detection, promoting adoption, validation, and collaborative improvements.