Related Experiment Video
Updated: Jun 14, 2026
04:54
IntelliSleepScorer, a Software Package with a Graphic User Interface for Mice Automated Sleep Stage Scoring
Published on: November 8, 2024
ActiTect: a generalizable machine learning pipeline for REM sleep behavior disorder screening through standardized
David Bertram1,2,3, Anja Ophey4,5, Sinah Röttgen5,6
1Faculty of Mathematics and Natural Sciences, University of Cologne, Cologne, Germany. dbertram@uni-koeln.de.
NPJ Digital Medicine
|June 12, 2026
Summary
ActiTect, an automated tool, accurately detects REM sleep behavior disorder (RBD) using wearable sensors. This open-source solution aids in early identification of synucleinopathies like Parkinson's disease.
Area of Science:
- Neurology
- Biomedical Engineering
- Data Science
Background:
- Isolated rapid eye movement sleep behavior disorder (iRBD) is a key prodromal marker for α-synucleinopathies, including Parkinson's disease and dementia with Lewy bodies.
- Wrist-worn actimeters offer potential for large-scale screening of RBD by monitoring nocturnal movements, but require robust analysis pipelines.
- Current methods for RBD detection from actigraphy data often lack generalizability across different devices and settings.
Purpose of the Study:
- To develop and validate ActiTect, a fully automated, open-source machine learning tool for identifying RBD from actigraphy recordings.
- To ensure the tool's generalizability across diverse acquisition settings through robust preprocessing and automated sleep-wake detection.
- To provide a reliable and efficient analysis pipeline for wearable-based RBD detection.
Main Methods:
- Development of a machine learning pipeline incorporating robust preprocessing and automated sleep-wake detection to harmonize multi-device data.
- Extraction of physiologically interpretable motion features from actigraphy recordings.
- Model training and validation on a multi-center cohort, including nested cross-validation and testing on independent external datasets.
Main Results:
- ActiTect demonstrated strong discrimination in identifying RBD, achieving an Area Under the Receiver Operating Characteristic curve (AUROC) of 0.95 in cross-validation.
- The tool showed consistent generalization performance on blinded local (AUROC=0.86) and external test sets (AUROC=0.84-0.94).
- Leave-one-dataset-out cross-validation confirmed robustness, with AUROC ranging from 0.84-0.89, and stability analysis indicated reproducible predictive features.
Conclusions:
- ActiTect provides a generalizable, automated, and open-source solution for detecting RBD using wearable actigraphy.
- The tool's robust performance across diverse datasets supports its potential for broader deployment in clinical screening and research.
- ActiTect facilitates independent validation and collaborative improvements, advancing wearable-based detection of prodromal synucleinopathies.
