JMIR Publications

ABSTRACT

Background:

Traditional clinical assessments in psychiatric research or clinical practice rely on global retrospective self-report depression measures, which do not adequately capture intra- and interpersonal variability in depressive symptoms over time and across contexts.

Objective:

This study aimed to (1) examine the feasibility of mobile ecological momentary assessment (EMA) for monitoring of depressive symptoms more sensitively than traditional scales; (2) identify subgroups at higher risk for depression and suicidal ideation and characterize their sociodemographic, psychological, and psychiatric profiles; and (3) determine whether repeated EMA-based self-monitoring in daily life reduces depressive symptoms.

Methods:

To capture the naturalistic severity and variability of depressive symptoms in everyday life, participants’ daily self-reports were collected via a mobile app–based EMA, indicating the presence or absence (1=yes, 0=no) of 20 depressive symptoms based on a 24-hour recall period, assessed once daily for 14 consecutive days. During this period, baseline questionnaires measured sociodemographic characteristics, digital sensitivity, and personality traits. On Day 14, postquestionnaires assessed state-trait anxiety, depression, insomnia, and somatic symptoms. In addition, a modified depression scale with a one-week recall period was administered on Day 7 and 14. For cluster analysis, three active EMA-derived features were included: mean symptom severity, within-person symptom variability, and frequency of suicidal ideation.

Results:

Compared to the 24-hour recall-based mobile EMA, the two-week recall-based traditional depression assessment significantly underestimated reported days of the nine major depressive episode (MDE) symptoms (all P<.001 for eight MDE symptoms; P<.05 for suicidal ideation). Cluster analysis of 695 participants (244 males and 451 females; age range 19–73 years, mean 36.14, SD 10.71) who completed at least seven EMA sessions over a 2-week period identified three distinct clusters: (1) no or low risk (n=445, 64.0%), (2) moderate risk (n=223, 32.1%), and (3) high risk (n=27, 3.9%). While severity of depressive symptoms, frequency of suicidal ideation, and all psychiatric characteristics progressively increased across clusters (cluster 1<2<3), the highest symptom variability was observed in cluster 2 (cluster 1<3<2). Notably, total scores obtained from the mobile EMA significantly decreased from the first to the second week (P<.001), suggesting that repeated self-monitoring may yield measurable improvements in depressive symptoms within a brief 2-week period.

Conclusions:

This large-scale community-based study demonstrated that mobile app–based EMA effectively reduced recall bias in assessing depressive symptoms and revealed potential therapeutic implications. Using a minimal set of active EMA-derived features to differentiate three distinct risk clusters, within-person symptom variability emerged as a clinically significant indicator not captured by traditional 2-week recall-based assessments. Despite the brief 2-week assessment period, these findings suggest that EMA can function not only as a robust data collection tool, but also as a low-intensity form of ecological momentary intervention.