JMIR Publications

ABSTRACT

Background:

Perinatal mental health disorders affect approximately 20% of women and are associated with substantial maternal and infant morbidity. Traditional assessment relies on infrequent, subjective self-reports. Mobile devices, including smartphones and wearables, offer continuous and objective measurement, but evidence on their assessment utility in perinatal populations remains fragmented.

Objective:

This study examines the application of wearable devices and smartphones for detecting and predicting perinatal mental health outcomes, with emphasis on predictive performance, informative features, and methodological rigor.

Methods:

We conducted a systematic review following PRISMA guidelines (PROSPERO: CRD420251249218). Six databases (PubMed, Web of Science, Scopus, PsycINFO, IEEE Xplore, ACM Digital Library) were searched in January 2026 without time restrictions. Evidence was synthesized narratively, and risk of bias was assessed using PROBAST+AI.

Results:

From 3,741 records identified across six databases, 10 studies met the inclusion criteria, covering postpartum depression (PPD), prenatal stress, discrete emotions during pregnancy (e.g., happiness, anxiety, sadness), and maternal loneliness. Models showed promising utility, particularly for PPD (multiclass AUC = 0.85; binary AUC = 0.871; F1 = 0.9872). HRV features (RMSSD, SDNN) were the most consistently informative physiological features, while GPS-derived mobility, physical activity, and sleep were the most informative behavioral markers, though their interpretation required perinatal-specific contextualisation. However, the field faces three critical constraints: methodologically, 67% of assessment units were rated at high risk of bias; in outcome scope, research has concentrated on PPD, while anxiety, stress, and other prevalent conditions remain largely unaddressed; and in technological breadth, sensing modalities and analytical approaches represent only a narrow subset of those now available.

Conclusions:

Wearable and smartphone sensing show early promise for predicting perinatal mental health outcomes, with autonomic and behavioral features emerging as complementary digital biomarkers whose interpretation requires perinatal-specific contextualisation. Advancing toward clinical utility will require broader coverage of mental health outcomes, larger longitudinal cohorts, standardised analytical pipelines and reporting practices, adoption of modelling approaches better suited to perinatal trajectories, and human-centred monitoring designs.