Digital Interventions for Depression in the AI Era: A Network Meta-Analysis to Compare and Rank Therapeutic Chatbots and AI-Augmented Therapy
Main Article Content
Abstract
INTRODUCTION: Artificial Intelligence (AI) is increasingly used in mental health care, with conversational agents offering scalable support for depression. However, the comparative effectiveness of AI-based therapies relative to other digital approaches remains unclear, and prior reviews have not provided treatment rankings.
OBJECTIVE: To compare and rank the effectiveness of AI-driven therapeutic chatbots (AI-DTA) versus non-AI digital interventions (NAI-DI) and control conditions in reducing depressive symptoms among adults.
METHOD: This network meta-analysis followed PRISMA-NMA guidelines. A systematic search across five databases (last updated September 17, 2025) identified randomized controlled trials involving adults (≥18 years) with Major Depressive Disorder or clinically significant depressive symptoms. The primary outcome was the standardized mean difference (SMD) in post-intervention depressive severity. Seven RCTs (n = 1,238) contributed to the connected evidence network. A frequentist random-effects NMA compared four intervention nodes. Transitivity was supported by comparable populations, delivery formats, and outcome timing. Heterogeneity was negligible (τ² = 0; I² = 0%; Q = 1.98, p = 0.37).
RESULTS: AI-DTA produced significantly greater reductions in depressive symptoms compared to NAI-DI (SMD = −0.55, 95% CI: −0.98 to −0.13). Differences for Sham and Information Control versus NAI-DI were not significant. AI-DTA ranked highest within the network (P-score = 0.943), and node-splitting analyses showed no significant inconsistency.
CONCLUSIONS: AI-driven therapeutic chatbots demonstrate superior effectiveness compared to non-AI digital interventions and may serve as a clinically useful adjunct or interim alternative when access to therapist-led care is limited. Further trials should evaluate long-term outcomes and direct comparisons with human-delivered therapy.
Article Details

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
© Journal of Asian Medical Students’ Association (JAMSA). Released under a Creative Commons license.