The Effects of Artificial Intelligence-Based Interventions on Depression and Anxiety: A Systematic Review

<ns4:p>Introduction Depression and anxiety remain major global mental health challenges that continue to increase across populations. Conventional treatments are often limited by cost, accessibility, stigma, and the availability of professionals. Artificial intelligence (AI)-based interventions have emerged as a potential approach to address these gaps. However, the growing body of evidence across diverse contexts calls for further synthesis. This study aims to examine research characteristics, evaluate effects, and analyse the implementation issues of AI-based interventions for depression and anxiety. Methods This systematic review was conducted in accordance with guidelines. Fourteen randomised controlled trials (RCTs) were identified from major databases, including Scopus, Web of Science, PubMed, and EBSCO, within the period from 6 November 2015 to 6 November 2025. Study quality was assessed using the Cochrane Risk of Bias 2 tool, and findings were synthesised using a narrative approach. Results The findings indicate that AI-based interventions, such as chatbots, large language models, and integrated platforms, generally demonstrate effects in reducing symptoms of depression and anxiety across various populations. However, results remain heterogeneous, with some studies showing outcome-specific or within-group improvements only. Implementation issues were identified, including limited human support, recruitment bias, and short follow-up periods, which may reduce adherence, generalisability, and the assessment of long-term effects. Conclusions AI-based interventions may be potentially accessible and scalable mental health solutions, with outcomes comparable to conventional care in certain contexts. However, their effects are shaped by implementation-related challenges, including variability in engagement, technological limitations, and ethical considerations. Future research should prioritise more standardised methodologies, longer intervention durations with follow-up, and greater attention to implementation design and sustainability. Systematic Review Registration Registered in PROSPERO on 16 February 2026 (Registration number CRD420261308648). Available from: https://www.crd.york.ac.uk/PROSPERO/view/CRD420261308648.</ns4:p>