JMIR Publications

ABSTRACT

Background:

Perinatal depression is a common and impactful condition affecting maternal and neonatal outcomes worldwide. Digital mental health interventions, particularly chatbots and conversational agents, have emerged as scalable and accessible tools to expand support across the perinatal continuum. However, the extent to which these technologies effectively support perinatal depression remains unclear.

Objective:

To map and synthesize the evidence on the use of chatbots and conversational agents for supporting perinatal depression, focusing on intervention characteristics, outcomes, safety mechanisms, and equity considerations.

Methods:

We conducted a scoping review following Joanna Briggs Institute methodology and reported according to PRISMA-ScR guidelines. Searches were performed in PubMed/MEDLINE, Scopus, Web of Science, CINAHL, Embase, LILACS, and Google Scholar up to April 2026. Eligible studies evaluated chatbot-based or conversational AI interventions providing psychological or emotional support to pregnant or postpartum women. Data were extracted on study design, intervention features, clinical and implementation outcomes, safety and screening mechanisms, and accessibility dimensions.

Results:

Nine studies were included, comprising one randomized controlled trial, multiple pilot and feasibility studies, one qualitative study, one real-world observational study, and one protocol. Most interventions demonstrated high feasibility, acceptability, and user engagement. The only randomized trial reported modest short-term reductions in depressive symptoms, with no consistent effects on anxiety or perinatal-specific measures. Substantial heterogeneity was observed in intervention design, theoretical frameworks, delivery platforms, and outcome metrics. Safety mechanisms, including automated risk detection and escalation, were limited or absent in most studies. Equity varied by delivery modality, with SMS-based interventions showing greater accessibility in low-resource settings, while app-based approaches required higher levels of digital access and literacy.

Conclusions:

Chatbots represent a promising and scalable approach to support perinatal mental health, particularly for expanding access to low-intensity psychological support. However, current evidence is limited by methodological heterogeneity, short follow-up, and insufficient safety integration. These tools should complement, rather than replace, professional care. Future research should prioritize rigorous trials, standardized outcomes, embedded safety protocols, and equity-oriented design to support integration into digital nursing care and health systems.