AI-informed mobile interventions to support adolescents’ mental health in schools: a pilot study

Introduction

Mobile apps provide a scalable approach to adolescent mental health support, and schools are well-placed to deliver large-scale interventions. By combining active (self-reported) and passive (mobile sensor) data, apps can model mental states and deliver context-aware support. MindCraft is an app integrating active and passive data to provide personalised AI-informed recommendations (“nudges”). We aimed to investigate the feasibility of delivering AI nudges via MindCraft to improve mental health outcomes among adolescents in UK schools.

Methods

The study was a three-arm pilot RCT. Following informed consent, participants completed a baseline assessment at school and downloaded the app. The primary outcome was the Strengths and Difficulties Questionnaire (SDQ). Secondary outcomes included eating disorders, insomnia, self-harm, self-efficacy, and well-being. Participants were randomised to: (1) personalised nudges (intervention), (2) non-personalised nudges (active control), or (3) self-monitoring (control), used the app for 4 weeks and were followed up at one month.

Results

375 students were recruited from 5 schools. After exclusions, 272 were included at baseline (mean age 16.88±1.34; SDQ 11.75±5.34). Follow-up data were available for 91 participants (66.5% attrition). 79.2% completed at least one active tracker, and 98.5% at least one nudge. ANCOVA showed lower SDQ scores in active control vs control (−2.84, SE 0.94, 95% CI − 4.70 to − 0.98, p = 0.003), with no difference between intervention and control.

Discussion

This study shows the feasibility of a school-based AI mental health trial. The study was underpowered and showed no statistically significant benefit of the nudging system of the intervention. A definitive trial is underway with improved retention strategies and inclusion of higher-risk participants, which will provide stronger evidence on mobile AI interventions for adolescent mental health and inform school-based prevention strategies.

Key messages

• Demonstrated the feasibility of implementing a school-based AI-enabled mobile intervention for adolescent mental health across multiple UK schools.

• Provided preliminary evidence that app-delivered nudges may improve adolescent mental health outcomes, supporting progression to a definitive full-powered randomised controlled trial.