Designing AI-powered healthcare assistants to effectively reach vulnerable populations with health care services: A discrete choice experiment among South African university students

Introduction South African young adults are at increased risk for HIV acquisition and other non-communicable diseases and face significant barriers to accessing healthcare services. The rapid development of artificial intelligence (AI), in particular AI-powered healthcare assistants (AIPHA), presents a unique opportunity to increase access to health information and linkage to healthcare services and providers. While successful implementation and uptake of such tools require understanding user preferences, limited understanding of these preferences exist. We sought to understand what preferences are important to university students in South Africa when engaging with a hypothetical AIPHA to access health information using a discrete choice experiment.

We conducted an unlabeled, forced choice discrete choice experiment among adult South African university students through Prolific Academic, an online research platform, in 2024. Each choice option described a hypothetical AIPHA using eight attribute characteristics (cost, confidentiality, security, healthcare topics, language, persona, access, services). Participants were presented with ten choice sets each comprised of two choice options and asked to choose between the two. A conditional logit model was used to evaluate preferences.

300 participants were recruited and enrolled. Most participants were Black, born in South Africa, heterosexual, working for a wage, and a mean age of 26.5 years (SD: 6.0). Results from the discrete choice experiment identified that language, security, and receiving personally tailored advice were the most important attributes for AIPHA. Participants strongly preferred the ability to communicate with the AIPHA in any South African language of their choosing instead of only English and to receive information about health topics specific to their context including information on clinics geographically near them. Results were consistent when stratified by sex and socioeconomic status.

Participants had strong preferences for security and language which is in line with previous studies where successful uptake and implementation of such health interventions clearly addressed these concerns. These results build the evidence base for how we might engage young adults in healthcare through technology effectively. Competing Interest Statement AZ LL CG CCM MPF SP declare no conflicts of interest. SM and DR work for Audere that develops health applications powered by artificial intelligence. Funding Statement The research and AZ were supported by an Established Investigator Award made from Boston University to LL. LL was supported by the National Institute of Mental Health of the National Institutes of Health under grant number K01MH119923. CG, CCM and SP were supported by the Bill and Melinda Gates Foundation under grant number AGMT DTD 8-4-23. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The content is solely the responsibility of the authors and does not necessarily represent the official views of Boston University, National Institutes of Health or the Bill and Melinda Gates Foundation. Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: The study protocol was reviewed and approved by the Human Research Ethics Committee of the University of Witwatersrand (231115) and the Boston University Medical Center Institutional Review Board (H-44483). Both approved a waiver of written informed consent so that participants could indicate consent online through an informed consent document shared with them on Qualtrics. The use of Prolific Academic meant that all participants were anonymous to the research team and no identifying information was provided. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes Footnotes Supported by: K01MH119923 (NIH) LL Established Investigator Award (Boston University) AZ AGMT DTD 8-4-23 (BMGF) CG CCM SP Data Availability Data is not publicly available but may be provided in a de-identified format upon reasonable request.