To monitor depression, a multidisciplinary team of researchers affiliated with SINTEF Digital, Haukeland University Hospital (NORMENT), the University of Bergen, the University of Oslo, and the Simula Metropolitan Center for Digital Engineering in Norway.
This repository presents a machine learning pipeline for classifying psychiatric patients versus healthy controls using wrist-worn actigraphy data.
The clinical data were collected within psychiatric research protocols and later made publicly available for machine learning analysis of motor activity patterns in depression.
Refer to the following link for the complete dataset:
🔗 https://datasets.simula.no/depresjon/
- Date Formating
- Behavioral feature engineering
- Leave-One-Subject-Out validation
- Logistic Regression
- Random Forest
The model achieves moderate classification performance, highlighting the potential and limitations of behavioral signals for psychiatric assessment.