An exploratory randomized controlled trial of an AI-enabled mental health intervention for generalized anxiety

Generalized Anxiety Disorder (GAD) is prevalent and often co-occurs with depression, contributing to significant disability and healthcare burden. Although treatments such as CBT and SSRIs are effective, access remains limited. This exploratory, randomized controlled trial evaluated the effectiveness of an AI-powered mental health app (PATH) in reducing symptoms of anxiety and depression. A total of 316 UK-based participants (aged 19-70) were randomized to either the intervention group (PATH) or a control group (NHS self-help website). The intervention provided evidence-based strategies including CBT-informed chat therapy and interactive tools. Anxiety (GAD-7) and depression (PHQ-9) scores were measured at baseline, two, eight, and twelve weeks. Of the 316 randomized participants, 235 completed the post-intervention assessment (33.0% attrition in the intervention group vs. 18.5% in the control group). Retention at 8-week and 12-week follow-ups was 77.4% and 54.0%, respectively. At two weeks, the intervention group showed significantly lower GAD-7 and PHQ-9 scores compared to control, with medium effect sizes. At eight weeks, those who continued using the app showed significant reductions in both anxiety and depression, while those who discontinued still showed medium improvements in anxiety. Effects were maintained at twelve weeks, with moderate to large effect sizes. Findings suggest that PATH significantly reduces anxiety and depressive symptoms, particularly with continued use. These results support the app's potential as a scalable, accessible digital intervention to address mental health treatment gaps.