JMIR Publications

ABSTRACT

Background:

Depression poses a global health challenge, and its early detection is critical for effective interventions. Recent studies reveal associations between digital traces of social behavior (e.g., phone calls) and depression, but rely on cross-sectional analyses, limiting insight into how these relationships evolve over time and obscuring the directionality between social behavior and mental health.

Objective:

This study investigates the longitudinal and directional relationships between digitally mediated social capital and depressive symptoms, leveraging phone call data to develop a data-driven framework for understanding depression over time.

Methods:

Eight weeks of data from 216 participants were analyzed using a dual-structural equation modeling (SEM) approach, including Latent Growth Curve Modeling (LGCM) and Cross-Lagged Panel Modeling. Digital social capital was operationalized through behavioral proxies capturing accessed social capital (e.g., incoming or outgoing calls) and latent social capital (e.g., missed phone calls), reflecting distinct mechanisms of social capital that are available and accessed online. Meanwhile, depressive symptoms were assessed using the Patient Health Questionnaire-4 (PHQ-4).

Results:

Latent growth analyses revealed that depressive symptoms were significantly associated with divergent trajectories of digital social capital. Higher baseline levels of depression were linked to changes in accessed social capital and declines in latent social capital. Growth in accessed social capital and depression were directly linked, indicating that increased levels of depression could amplify how often latent social capital becomes accessed. Cross-lagged panel analyses further corroborated these findings by showing that depressive symptoms in the beginning of the study were related to subsequent reductions in latent social capital, whereas prior levels of latent social capital were not significantly correlated to higher levels of depression.

Conclusions:

These findings advance the clinical understanding of depression by revealing that psychological health could actively influence patterns of social engagement. They also suggest that changes in social behavior could reflect changes in depressive symptoms. This study highlights the importance of longitudinal, data-driven approaches for interpreting digital social traces and underscores their potential for informing mental health scholarship and intervention strategies.