AI-Based Approaches to Refugee Mental Health Care: Systematic Integrative Review

Refugees and forcibly displaced populations experience elevated rates of mental health conditions, including posttraumatic stress disorder, depression, and anxiety, while facing substantial barriers to mental health care. AI has emerged as a promising approach for mental health detection, intervention, and decision support; however, no review has specifically examined AI-based approaches to refugee mental health care.

This systematic integrative review aimed to (1) examine how AI has been used in refugee mental health research and care, (2) synthesize the reported strengths and limitations of AI-based approaches, and (3) identify future directions for this field.

An integrative review methodology was used following Whittemore and Knafl's framework. Systematic searches were conducted across PubMed, Embase, CINAHL, PsycINFO, Scopus, Web of Science, Cochrane Library, ACM Guide to Computing Literature, IEEE Xplore, arXiv, and medRxiv from inception through June 2026. The first 100 results from Google Scholar sorted by relevance were additionally screened. Studies were included if they focused on refugee populations and applied AI technologies to refugee mental health care or assessment, while studies focusing on voluntary migrants and digital mental health not using AI were excluded. Quality appraisal was performed mainly using the Mixed Methods Appraisal Tool. Data were synthesized in accordance with the integrative review methodology.

Of 2726 studies identified, 23 studies were included in the review. AI-based approaches were classified into 3 functional categories: Detection, Prediction, and Classification (15/23, 65.2%); Intervention and Clinical Support (4/23, 17.4%); and Information Synthesis (5/23, 21.7%). Reported strengths of AI-based approaches included scalability, accessibility in resource-limited settings, support for early symptom detection, reduction of language barriers, and efficient synthesis of large unstructured data. Common limitations included potential biases, lack of contextual understanding, structural barriers in technology access, and concerns regarding overreliance on AI-based systems. Future directions emphasized culturally sensitive model development, larger, multimodal datasets, clinical validation, human-centered design, and implementation research.

Unlike previous reviews that focused primarily on digital mental health interventions or AI applications in general populations, this review provides the first comprehensive synthesis of AI-based approaches specifically for refugee mental health care. This review identified 3 functional domains that characterize the current landscape of AI applications in refugee mental health care and highlights key methodological, ethical, and implementation considerations. Despite heterogeneity in the included studies, the findings provide a foundation for future research and may inform the responsible development, evaluation, and implementation of AI-based approaches in refugee mental health care across humanitarian settings.

PROSPERO CRD420251246982; https://www.crd.york.ac.uk/PROSPERO/view/CRD420251246982.