Enhancing the accuracy of mental health assessments through the integration of self-report and objective measures: A convergence study utilizing biosignals and 14-day wearable data

Abstract

Traditional mental health assessments have primarily relied on self-report surveys. With the advancement of biosignal and daily life data acquisition technologies, there is growing potential to enhance the accuracy of mental health evaluations. This study examined the predictive value of self-reported and objective measures in assessing depressive symptoms and evaluated whether their integration improves model performance. Forty-three police officers completed standardized mental health questionnaires, including the Patient Health Questionnaire-9 (PHQ-9), Generalized Anxiety Disorder-7 (GAD-7), and Brief Resilience Scale (BRS). Participants performed a laboratory-based mental arithmetic task designed to induce acute stress while their electrocardiogram (ECG) was recorded to extract heart rate variability (HRV) features. Additionally, they wore a smartwatch for 14 consecutive days to monitor stress levels multiple times daily, as well as continuous sleep and activity patterns. Features were extracted from psychological, physiological, and daily life data. Hierarchical regression analyses revealed that the baseline model including demographic variables explained 9.5 % of the variance in depressive symptoms. Adding psychological measures increased the adjusted R2 to 0.478 (ΔR2 = 0.380, p < .001). Including HRV features led to a modest increase (adjusted R2 = 0.505; ΔR2 = 0.047, p = .152). The final model, which integrated wearable-derived stress and sleep variables, significantly improved predictive accuracy (adjusted R2 = 0.700; ΔR2 = 0.199, p = .001). These findings suggest that while self-report assessments remain critical, integrating multimodal data from wearable devices can substantially enhance the prediction of depressive symptoms, particularly in shift-working populations such as police officers.