This page was translated automatically from Turkish. Read the original

This page is an AI-assisted Turkish summary of the study presented at the 26th World Congress of Psychiatry. The study belongs to the presenting authors; please refer to the original source when citing it.

SO123

AI-based mental health chatbots for adolescents: a systematic review of engagement, acceptability, and therapeutic outcomes

AI-BASED MENTAL HEALTH CHATBOTS FOR ADOLESCENTS: A SYSTEMATIC REVIEW OF ENGAGEMENT, ACCEPTABILITY, AND THERAPEUTIC OUTCOMES

AI-based chatbots appear feasible and acceptable for adolescents, but evidence of therapeutic benefit is preliminary and limited to three small studies.

This systematic review aimed to examine recent empirical studies evaluating engagement, acceptability, safety, and therapeutic outcomes of artificial intelligence (AI)-based mental health chatbots in adolescents; the abstract specifies an age range of 10–19 years. PubMed, EBSCO Open Dissertations, and ScienceDirect were searched in January 2026 for studies from the preceding five years; in the PRISMA flow diagram, of 556 records, 209 were screened after records were removed using database filters (n=342) and duplicates were excluded (n=5). Twenty-two full texts were assessed, 14 were excluded for lacking adolescent-specific data, and 5 for low methodological quality, leaving 3 included studies with a total of 196 participants. The included studies comprised a secondary analysis of a quasi-randomized trial (#LIFEGOALS, 12 weeks, n=159), a single-group pre–post pilot study (REALbot, 1 week, n=20), and a pilot randomized controlled trial (W-GenZ, 4–12 weeks, analyzed n=17). Regarding engagement, 76.1% of #LIFEGOALS users did not reopen the app during weeks 7–12; more severe depressive symptoms were associated with lower app use (OR 0.33; 95% CI 0.17 to 0.66; p=0.002), whereas lower peer support was associated with higher use (OR 0.43; 95% CI 0.25 to 0.74; p=0.002). Acceptability and usability were high: 8 in 10 adolescents liked W-GenZ, and REALbot received a usability score of 78 out of 100. Regarding therapeutic outcomes, no significant clinical change was observed with REALbot (social isolation from 30.2 to 28.8, p=0.44; depression from 14.3 to 11.6, p=0.08); in the pilot randomized trial, PHQ-9 scores decreased by 3.3 points in the chatbot arm and 2.0 points in the waitlist control arm at week 4, but the pilot study was not powered to demonstrate superiority. Regarding safety, suicidal ideation alerts were triggered for 10 of 17 participants across the two groups, with 4 psychiatric assessments, 2 referrals, and 1 emergency department visit; no causal link to the intervention was established. The authors recommend considering chatbots as adjuncts to care and conducting larger randomized trials with longer follow-up to evaluate safety, privacy, and design elements.

All WCP 2026 summaries