JMIR Publications

ABSTRACT

Background:

Given the increasing prevalence of mental health problems among adolescents, early intervention and appropriate management are needed to decrease mortality and morbidity. Artificial Intelligence (AI) 's potential contributions, although significant in the field of medicine, have not been adequately studied in the context of adolescents’ mental health.

Objective:

This review therefore aimed to identify AI interventions that have been tested and/or implemented for use in adolescents' mental healthcare.

Methods:

We used Levac et al. and the Joanna Briggs Institute scoping review framework for conducting this review. We followed PRISMA-ScR (Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews) checklist for reporting the study. We searched five electronic databases from inception date until July 2024. Two independent reviewers screened the articles and extracted the data. A third reviewer resolved disagreements. We evaluated the risk of bias for prognosis and diagnosis-related studies using Prediction model Risk Of Bias Assessment Tool (PROBAST).

Results:

We retrieved 4213 records, and 88 papers were included after the two-level screening. In the included papers, AI was used for diagnostic (n=78), monitoring and evaluation (n=19), treatment (n=10), and prognosis (n=6). For diagnosis purposes, the papers focused on: autism spectrum disorder (n=7), followed by suicidal behaviours (n=11). Machine learning methods were the most frequently reported AI methods. The "overall" risk of bias for diagnosis/prognosis models was unclear (58%).

Conclusions:

Overall, within our included papers, AI was utilized to support various aspects of adolescent mental healthcare, including diagnosis, treatment, monitoring, and prognosis. However, there was a notable emphasis on diagnosis in the available literature, with limited exploration of AI applications in other areas. This research gap should be addressed in future studies. Moreover, future studies should address work on meaningful and active involvement of end-users in designing, developing, and validating AI interventions, as well as better reporting of AI models, data collection, and analysis process.