Integrating Artificial Intelligence into Psychosocial Support for Anxiety and Stress Among People Living with HIV: A Scoping Review and Evidence Map with Implications for Newly Diagnosed Populations in Nigeria

Abstract

Background: The period surrounding an HIV diagnosis may be accompanied by anxiety, psychological distress, stigma, disclosure concerns and uncertainty about treatment and relationships. In many resource-constrained services, brief counselling remains the principal immediate response, while specialist mental health support is limited. Artificial intelligence-enabled systems and wider digital interventions may extend psychosocial support, but their clinical effectiveness, cultural fit and governance requirements remain uncertain. Objective: To map verified evidence on AI-enabled and digitally mediated psychosocial interventions relevant to anxiety and stress among people living with HIV, examine the evidence for the immediate post-diagnosis period, and identify implications for research and implementation in Nigeria. Methods: The review was structured using established scoping-review frameworks and PRISMA-ScR guidance. A standardised charting form captured study design, population, intervention, comparator, duration, psychosocial outcomes, engagement, principal findings and limitations. Interventions were classified as AI-enabled, rule-based or adaptive, or digitally mediated without AI. A reconstructed evidence set of 17 verified publications from 2017 to 2025 was synthesised narratively. Results: The 17 publications represented nine countries and 16 parent intervention programmes; ten were randomised or pilot randomised trials. Only two publications clearly evaluated AI-enabled systems, three described rule-based, automated or technically unclear systems, and 12 evaluated non-AI digital interventions. Selected non-AI interventions improved anxiety, depression, stigma, coping or self-management, although findings were inconsistent and many studies prioritised feasibility or engagement. Neither AI study established psychosocial effectiveness. No publication specifically evaluated prevention of anxiety disorders or sustained stress reduction during the immediate post-diagnosis period. A Nigerian single-arm study reported fewer depressive symptoms after 48 weeks of SMS reminders plus peer navigation, but it was not AI-enabled and causal attribution was limited. Conclusion: AI-enabled and other digital interventions may complement HIV care, but current evidence does not establish that AI prevents anxiety disorders or meaningfully reduces stress among newly diagnosed people living with HIV. Nigerian development should begin with culturally co-designed, clinician-supervised, low-data interventions integrated with existing HIV and mental health services, supported by validated outcomes, crisis escalation, equitable access and strong data governance.