618. A RANDOMISED CONTROLLED TRIAL OF AN AI-ENABLED MENTAL HEALTH INTERVENTION FOR GENERALIZED ANXIETY DISORDER SYMPTOMS

Abstract

Background: Generalized Anxiety Disorder (GAD) is a prevalent mental health condition, often co-occurring with major depression and contributing to disability and healthcare costs. While treatments like Cognitive Behavioural Therapy (CBT) and SSRIs are effective, access is limited by provider shortages, cost and stigma. AI-powered digital interventions offer a scalable alternative. Aims & Objectives: This randomized controlled trial (RCT) aims to assess the impact of Dr Jay app (PATH) in reducing symptoms of generalized anxiety, as well as co-occurring depressive symptoms. Method: A total of 316 participants (201 female and 115 male), aged 19 to 70 and residing in the UK, were randomly assigned to intervention (PATH) or control groups (NHS website featuring selfhelp resources). Participants in the intervention group engaged with an app that delivered evidence-based strategies, including CBT-based chat therapy, self-tests, and interactive exercises. After twoweeks, participants in the intervention group either continued or discontinued using the app to enable additional subgroup analysis. Control group participants were instructed to use the NHS self-help resources. Anxiety (GAD-7), depression (PHQ-9) scores were measured at three time points: baseline, two weeks, and eight weeks. Mixed ANOVA and t-tests were used to assess group differences and effect sizes. Results: At two weeks, the intervention group had significantly lower GAD-7 scores (M = 11.43, SD = 4.64) compared to the control group (M = 12.85, SD = 4.26), reflecting a medium effect size (p = .040, d = 0.32). PHQ-9 scores were also significantly lower in the intervention group (M = 13.16, SD = 5.39) compared to the control group (M = 14.95, SD = 5.20), with a medium effect size (p = .031, d = 0.34). At eight weeks, GAD-7 scores of intervention participantswho discontinued using the app after twoweekswere significantly lower than those of the control group (p = .024, d = 0.43), with medium effect size. GAD-7 scores of participants who continued using the application (M = 6.87, SD = 5.22) were also lower than those of the control group (M = 12.46, SD = 4.67), ref lecting a large effect size (p < .001, d = 1.18). At eight weeks, PHQ-9 scores of participantswho discontinued using the app after twoweeks (M = 12.36,SD = 6.58)were lower than those of the control group but the difference did not reach statistical significance; scores of participants who continued using the application (M = 9.60, SD = 6.75) were significantly lower than those of the control group (M= 14.29, SD = 5.56), reflecting a large effect size (p = .005, d = 0.82). Discussion & Conclusions: The findings provide strong evidence for the effectiveness of AI-driven mental health interventions. PATH significantly reduced symptoms of anxiety and depression, with continued users experiencing greater benefits. These results highlight its potential as a scalable solution to bridge gaps in mental healthcare.