JMIR Publications

ABSTRACT

Background:

Anxiety symptoms during public health crises are associated with adverse psychiatric outcomes and impaired health decision-making. The interaction between real-time social media use patterns and clinical anxiety during infectious disease outbreaks is underexplored.

Objective:

This work aimed to evaluate the usage pattern of two types of social media apps (communication and social network) among patients in outpatient psychiatric treatment during the COVID-19 surge and lockdown in Madrid, Spain, and their short-term anxiety symptoms (GAD-7) at clinical follow-up.

Methods:

The individual-level shifts in median social media usage behavior from Feb 1st through May 3rd, 2020were summarized using repeated measures analysis of variance (ANOVA), accounting for the fixed effects of lockdown (pre-lockdown versus post-lockdown), group (clinical anxiety group versus non-clinical anxiety group), the interaction of lockdown and group, and random effects of users. A machine learning (ML)-based approach combining a hidden Markov model (HMM) and logistic regression was applied to predict clinical anxiety group (n=44) from non-clinical anxiety group (n=51), based on the longitudinal time-series data of communication and social network app usage (in seconds) as well as anxiety-associated static variables that showed significant differences between the groups, including the presence of an essential worker in the household, worries about life instability, and health status.

Results:

Individual-level analyses of daily social media usage showed that an increase in communication app usage from pre-lockdown to lockdown period was significantly smaller in the clinical anxiety group (n=37) than in the non-clinical anxiety group (n=37), F1, 72 3.84, p=0.05. The ML model achieved 61.80% mean accuracy and 0.70 area under the receiver operating curve in predicting the clinical anxiety group from high social network and low communication app usage.

Conclusions:

Patients who reported fewer anxiety symptoms were more active in communication apps after the mandated lockdown and less active in social network apps in the overall period, suggesting a differential pattern of digital social behavior in adapting to the crisis. Passive-sensing of a shift in category-based social media app usage during the lockdown can predictively model digital biomarkers moderating the severity of anxiety symptoms.