Integrating artificial intelligence and machine learning in HIV testing interventions in Gauteng Province, South Africa: Opportunities, challenges, and implementation strategies

Abstract

Background: Conventional HIV testing approaches continue to fall short of overcoming barriers to HIV testing, especially among key and priority populations at higher risk of acquiring and transmitting HIV. Artificial intelligence (AI) and machine learning present a unique opportunity to strengthen prioritised HIV testing through risk prediction and enhanced diagnostic tools. Objective: This study discussed stakeholders' views on opportunities, challenges, contextual considerations and an implementation roadmap and strategic recommendations for integrating AI and machine learning into HIV testing in South Africa. Method: This qualitative study recruited 15 stakeholders in Gauteng Province, using individual semi-structured face-to-face interviews. Thematic content analysis was performed, and the Consolidated Framework for Implementation Research was used to map the implementation roadmap of the results. Results: Four superordinate themes were identified: perceived benefits, challenges, ethical considerations and implementation strategies. The study discussed the opportunity to leverage AI to enhance HIV testing through HIV risk prediction, self-testing support and advanced, accurate diagnostics. However, technological access, digital divide, resource constraints, privacy concerns, skill gaps and staff resistance, among other barriers, were noted. Conclusion: The implementation design should incorporate the perspectives of all stakeholders involved in HIV testing to address human factors and ethical concerns surrounding AI use.