An Umbrella Review of Artificial Intelligence Applications in Mental Health Care

Vollständiger Abstract

Worum geht es in dieser Arbeit?

ABSTRACT Background Artificial intelligence (AI) is increasingly used in mental health care to address rising demand, workforce shortages and access barriers; however, evidence remains scattered across multiple systematic reviews, limiting synthesis and practical application. Objective To synthesise evidence on AI applications in mental health care, including trends, uses, benefits, challenges and risk‐mitigation strategies, guided by an ethics of care framework. Methods This umbrella review of systematic reviews followed the PRISMA 2020 guidelines. Five databases (Web of Science, Scopus, PubMed, PsycINFO and CINAHL) were searched for peer‐reviewed studies published between 2021 and 2025, with additional studies identified through backward and forward citation tracking. Eligible studies were systematic reviews examining artificial intelligence (AI) applications in mental health care involving human participants. Exclusion criteria included non‐systematic reviews, non‐AI studies, grey literature and non‐English publications. Three reviewers independently screened titles, abstracts and full texts according to predefined inclusion and exclusion criteria. Methodological quality was assessed using the AMSTAR‐2 tool, and data were extracted on AI type, mental health focus, end‐user applications, benefits and challenges. Extracted data were analysed thematically using an iterative coding process to identify recurring patterns on AI type, mental health focus, applications, benefits and challenges. Results Twenty‐seven systematic reviews with over 14 million participants were included. Findings converged across four primary themes. First, AI enhances early detection and risk stratification, particularly for depression, anxiety, stress, PTSD and suicidal ideation. Second, AI‐enabled digital tools, such as chatbots and mobile platforms, expand access and engagement by supporting symptom monitoring, self‐management and continuity of care. Third, structural and ethical fragilities were noted, including data bias, limited external validity, methodological heterogeneity and insufficient transparency. Fourth, AI demonstrated greatest value when embedded within human‐centred, hybrid care models that preserve clinical judgement, relational care and accountability. Evidence supporting AI applications was stronger for screening and predictive tasks than for sustained therapeutic interventions. Among the 54 reviews assessed for methodological quality, 27 achieved moderate or high confidence ratings on the AMSTAR‐2 tool, with stronger evidence observed in reviews evaluating AI interventions for early detection and risk prediction. Conclusion Artificial intelligence (AI) has significant potential to enhance mental health nursing by supporting early identification of symptoms, improving access to care and strengthening clinical decision‐making. For nursing practice, responsible implementation requires rigorous validation, equity‐focused design, transparency and sustained human oversight. Integrating AI within an ethics‐of‐care framework ensures that these technologies support person‐centred, relational nursing care rather than replacing the critical role of nurses in therapeutic engagement and clinical judgement.