Acceptability, Usability, and Clinical Appropriateness of an AI-Powered Tool for Linkage to HIV Services After HIV Self-Testing in South Africa: Mixed Methods Evaluation

Abstract Background Timely linkage to HIV prevention and treatment services following HIV self-testing (HIVST) remains a challenge in many countries. While HIVST offers privacy and convenience for individuals to know their HIV status, many do not engage in follow-up care due to behavioral barriers like uncertainty about next steps, stigma, and privacy concerns. Objective This study evaluated the acceptability, usability, and clinical appropriateness of Your Path, an AI-powered tool designed to facilitate access to HIV services following HIVST. Methods This study was conducted with community members and health care providers, recruited from Indlela’s Behavioral Hub in South Africa. Community members were randomly assigned a mock HIV result and completed a pretest survey assessing intentions to seek HIV services, followed by a simulated HIVST guided by Your Path , delivered via WhatsApp (Meta Platforms). The tool provides guidance for HIVST support, interpretation of results, emotional support, and delivers tailored information to access HIV services through an interactive chat interface. Community members then completed a posttest survey assessing intentions to seek HIV services and usability of the tool using the System Usability Scale (SUS) questions. A subsample completed in-depth interviews exploring their experience with Your Path. Health care providers reviewed interaction transcripts, assessing clinical appropriateness, completeness, and relevance of the summaries; qualitative data underwent thematic analysis using the Theoretical Framework of Acceptability. Results A sample of 100 community members aged between 18 and 61 years was enrolled: 59 (59%) were female, and 51 (51%) were assigned a negative mock HIV result. Overall, the intervention had its strongest effect on participants’ confidence in accessing HIV services, particularly among those with HIV-negative results (Wilcoxon P =<.001; Stuart-Maxwell P =.004). Intentions and perceived importance were high at baseline and remained largely unchanged over time in both groups. The tool demonstrated high usability, with a mean SUS score of 81.6 (SD 17.5) out of 100. Most found Your Path easy to use (84/100, 84%), 76 of 100 (76%) did not need assistance, and 95 of 100 (95%) expressed willingness to use it again. However, 15 of 100 (15%) noted it did not always provide consistent responses to questions. From qualitative interviews, the majority described Your Path as helpful, user-friendly, acceptable, and private. Participants appreciated its private and nonjudgmental tone and confidentiality, though some expressed information protection concerns. Twenty-five health care providers aged between 24 and 64 years participated in qualitative interviews. The majority found the summaries clinically appropriate, informative, and beneficial for HIV care linkage, but noted limited emotional responsiveness and language barriers. Conclusions Your Path was acceptable, usable, and clinically appropriate for supporting access to HIV services after HIVST. Larger in-field studies are needed to understand the implementation, cost, and sustainability of digital interventions in public health.