AI-enabled and comparator digital interventions for adolescent mental health (2020–2025): early detection, human–AI collaboration, and ethical governance—a PRISMA-ScR scoping review

Abstract

Mental health difficulties among adolescents and youth (ages 10–24) are increasing globally, yet early intervention remains impeded by stigma, service shortages, and cultural barriers. AI-enabled technologies including chatbots, automated screening tools, and machine learning-based predictive models have been proposed as scalable approaches to address these gaps, alongside non-AI digital interventions such as therapist-guided internet-delivered cognitive behavioral therapy (iCBT) and smartphone-based CBT applications, which are examined as comparator evidence given their stronger controlled-design evidence base. This scoping review followed Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) reporting guidelines and Joanna Briggs Institute (JBI) methodology. A systematic database search across six databases (PubMed, Scopus, ProQuest, SpringerLink, Web of Science, and ScienceDirect), supplemented by manual and forward-citation searching, yielded 10,962 records, of which 19 studies met the predefined inclusion criteria. A Human-AI Collaboration Role framework comprising a four-level continuum from fully human-led to fully automated delivery was applied as a predefined coding dimension across all included studies. Most studies focused on depression and anxiety among mid-to-late adolescents in high-income countries (68.4%). Therapist-guided iCBT was the most rigorously studied format using randomized and controlled designs; AI chatbots, predictive models, and large language model (LLM)-based systems demonstrated feasibility or engagement outcomes in a limited number of studies and contexts, though randomized diagnostic chatbot trials specifically remain absent for participants under 18 years; at least one randomized non-diagnostic chatbot trial (a body-image intervention) has been conducted in this age group. Equity-related reporting was inconsistent: socioeconomic status was reported in only 10.5% of studies and cultural adaptation in 42.1%; restriction to English-language publications may further underrepresent non-Western and low- and middle-income-country evidence relevant to these equity findings. Key implementation challenges included inconsistent safeguarding protocols for minors, algorithmic bias, data privacy concerns, and insufficient integration with clinical and educational systems. The evidence base remains preliminary and methodologically heterogeneous, insufficient to support conclusions about clinical effectiveness or generalizability to diverse adolescent populations. These findings represent a descriptive map of the current evidence landscape rather than an assessment of effectiveness. Future research should prioritize adequately powered adolescent-focused trials, participatory co-design approaches, culturally informed adaptations, and transparent ethical governance frameworks to ensure responsible development of AI-enabled digital mental health tools for youth. Because the included studies were heterogeneous and no formal risk-of-bias assessment was undertaken, this review does not establish comparative clinical effectiveness, safety, or superiority of AI-enabled interventions. https://osf.io/sbpk2 , doi: 10.17605/OSF.IO/SBPK2 .