Wearable-Derived Evening Motion Entropy and Morning Light Exposure Are Associated with Depressive Symptoms in University Students.

Abstract

Depressive symptoms are prevalent in university students, yet scalable screening is constrained by self-report and intermittent assessment. Wearable monitoring of motion and light offers an objective alternative, though their combined screening characterization is limited. Undergraduate volunteers wore a self-designed wristband for seven days, recording triaxial acceleration and ambient light at 1 min resolution. Depressive symptoms were assessed via nine-item Patient Health Questionnaire (PHQ-9) and grouped as Dep (≥5) or HC (<5). Motion and light rhythm features were extracted from 24 h profiles and predefined windows; logistic regression (LR) and support vector machine (SVM) compared motion-only, light-only, and joint sets, interpreted with Shapley additive explanations (SHAP). After preprocessing, 152 participants provided valid activity data (HC = 132, Dep = 20) and 218 valid light data (HC = 186, Dep = 32). The Dep group showed delayed activity timing, lower daily light exposure, and greater evening motion irregularity, with differences most prominent during 18:00-22:00 for motion entropy and variance. The joint motion-light SVM performed best (accuracy 85.2%, recall 60.0%, F1 Score 46.2%, Area Under Curve 0.892). SHAP highlighted evening motion entropy and variance, morning light exposure, and rhythm-related wavelet features. These findings indicate that wearable motion and light rhythms are associated with depressive symptoms; evening motion entropy and morning light exposure may aid campus risk screening as interpretable markers, although the cross-sectional design and PHQ-9-based grouping preclude diagnostic or causal conclusions.