214 - OSMe (“Awesome”) Buddy: An AI Chatbot Pilot Intervention to Promote HIV Prevention and Navigation for Latino Youth and Young Adults

Publication Number: 2206.214

Harita Shah, University of Chicago Division of the Biological Sciences The Pritzker School of Medicine, Chicago, IL, United States; Pedro A. Serrano, CQLC NFP, Chicago, IL, United States; Khader Alisha Bin Ahmed Mohammed, Versante, Carmel, IN, United States; Gregory Phillips, Northwestern University The Feinberg School of Medicine, Chicago, IL, United States

Background: HIV disproportionately impacts Latino/e sexual and gender minority individuals, particularly youth and young adults. In order to promote HIV prevention, our community-academic partnership developed the PrEPárate (“Be PrEPared”) social marketing campaign in 2022. The campaign featured a bilingual website and hotline to connect individuals to pre-exposure prophylaxis (PrEP) information and local resources. However, many HIV prevention questions arise outside of hotline staffing hours. To extend capacity to attend time-sensitive health questions during unstaffed hours, we are piloting an artificial intelligence (AI) chat-based assistant for the PrEPárate website called the OSMe (Outreach, Screening, and Management) Buddy (pronouned "Awesome Buddy"). OSMe Buddy is a generative conversational AI designed for health navigation through stigma-free pathways to care, thereby addressing social determinants of health to improve the health of marginalized communities.

Objective: (1) To describe community-driven implementation of an AI chatbot intervention for HIV prevention
(2) To study engagement patterns with an AI chatbot pilot tailored for young Latino/e populations accessing health information

Design/Methods: We piloted an AI chatbot for HIV prevention and navigation through a multidisciplinary partnership between the Chicago Queer Latine Collaborative, the University of Chicago, and Versante LLC. We adapted the OSMe Buddy AI chatbot for use in our PrEP campaign website, ensuring its proficiency in English and Spanish, and refining its database of Chicago-based HIV resources. To understand how the chatbot is being used and by whom, we designed a dashboard to collect metrics of user engagement. We present formative research on community-engaged approaches to AI implementation and descriptive statistics of the intervention's pilot phase.

Results: In the first 2 months of the pilot phase (Aug 29 – Oct 29, 2025), 308 users engaged with the chatbot. 64% inquired about PrEP and HIV services, and 36% engaged in light conversation. The majority of engagement occurred between 11am and 4pm. Additional dashboard data will be forthcoming including socio-demoigraphics of users, language of access, chat content, and referral outcomes.

Conclusion(s): This initiative demonstrates a novel model for empowering frontline community organizations to lead the adoption, ethical co-design, and rigorous evaluation of AI tools. By centering community leadership and ensuring cultural competence, this approach maximizes the potential of AI to reduce health disparities among marginalized populations and catalyze health behavior change.