Using our well-established research infrastructure, we will recruit key stakeholders nationwide who may contribute to understanding AI-powered chatbots in the HIV care and prevention continuum using semistructured in-depth interviews. These key stakeholders include PrEP-eligible Black cisgender women (n=50: n=10, 20% in phase 1; n=20, 40% in phase 2; and n=20, 40% in phase 3) and NPs who provide HIV treatment and prevention services to patients (n=10). We will use semistructured interviews to elicit information regarding these stakeholders’ opinions on chatbot development and evaluation. Black cisgender women who participate in phase 1 will not be invited to phases 2 and 3 to avoid the issue of information contamination. A community advisory board (CAB) with volunteer members (n=10) representing key stakeholders will be established to ensure our research reflects the priorities of the study population. These members will be selected to represent different characteristics (eg, socioeconomic status and geolocations) in the targeted community. Leaders from local community organizations and general community members will also be invited to strengthen the acceptability and quality of the proposed research. Once the CAB members are recruited, they will remain throughout the study.

For PrEP-eligible Black cisgender women, we will recruit 50 Black cisgender women who are eligible for HIV prevention services. The inclusion criteria include (1) aged 18 years or older, (2) self-identifying as Black or African American, (3) self-reported as female, (4) not tested as HIV positive (either HIV negative or unknown status), (5) eligible for HIV prevention (engaging in unprotected sex, having multiple sexual partners, or history of sharing needles with others), (6) willing to provide consent for participation, (7) fluent in English, and (8) able to engage in either in-person or web-based interviews. For HIV care providers, we will recruit 10 NPs who provide HIV prevention or treatment to patients. Eligible health providers need to (1) be aged 18 years or older, (2) work as a nurse or NP, (3) have ever provided HIV prevention or treatment in the past 12 months, (4) be willing to provide consent for participation, (5) be fluent in English, and (6) be able to engage in either in-person or web-based interviews. For CAB members, we will invite 10 members representing different characteristics (eg, socioeconomic status and geolocations) in the targeted community. Eligible CAB members need to be (1) aged 18 years or older, (2) interested in HIV prevention and treatment topics, (3) fluent in English, (4) willing to provide consent for participation, and (5) able to engage in either in-person or web-based meetings.

Participants will be recruited via multiple routes. For HIV nurses or NPs, an institutional review board (IRB)–approved recruitment message and flyers will be distributed via local newsletters or the listserve among members of the Association of Nurses in AIDS Care or the National Black Nurses Association using the IRB-approved recruitment email script message for nurses. We will recruit CAB members via email listserves and flyers for recruitment. For PrEP-eligible Black cisgender women, we will distribute the recruitment information via the regional ResearchMatch portal to recruit willing volunteers. Anyone interested in this study can use the survey link or scan the QR code for eligibility screening. They can also contact the research team members via email or phone, and the research team will direct them to the screening procedures. Other interested participants will be recruited using “snowball” chain referral sampling, where we ask Black cisgender women, nurses, or potential CAB members to identify others within their networks who might be interested in the study. Participants can either complete the brief screening over the phone or be emailed a link to complete the screening via REDCap (Research Electronic Data Capture; Vanderbilt University). A referrer will be compensated a small amount (ie, US $5) for each peer they successfully refer, with a cap on referral payments at 3 successful enrollments from participants. When potential participants contact us, the study team members will explain the study and obtain verbal consent to proceed with phone screening using the script, or they can access the REDCap link directly from the study flyer by scanning the QR code.

In this proposed study, our primary objectives will be achieved through the following 3 phases within the 18-month timeframe by recruiting 50 Black cisgender women and 10 NPs.

Quantitative data include demographic information, socioeconomic characteristics, and implementation and usability outcomes. Due to the limited sample size (n=20), we will perform descriptive analyses to present participants’ characteristics (eg, demographics, sexual behaviors, and HIV risks) and user experiences (eg, feasibility and acceptability) to observe potential similarities and differences in particular characteristics’ implementation and use outcomes (eg, age group). Qualitative data (ie, opinions regarding the chatbot development and experience) will be transcribed and loaded into qualitative analytical software for analysis. This will follow the 5-step framework analysis approach, which includes familiarization (describing potential themes), thematic framework (generating a nascent code or subcode list), indexing (producing a final consensus code list), charting (organizing critical thematic constructs into hierarchical themes), and mapping and interpretation (summarizing and interpreting) [32]. Participants’ verbal utterances collected by think-aloud sessions will be transcribed and analyzed by propositional analysis (an analysis to create a basic semantic model of the text to comprehend an individual’s thoughts and critical ideas for a given product) [33]. Usability data gathered by the dashboard will be used to describe user activities and profiles, such as log-in time, bounce rate, retention rate, and conversation channels. Conversation data will be downloaded from the dashboard and explored using the systematic thematic analysis strategies (familiarization, code assignment, pattern identification, and defining and reporting themes) [34-36]. Specifically for data collected from the NPs, we will analyze both quantitative and qualitative responses. Quantitative data will include NPs’ demographic information, professional background, and their engagement with PrEP care. We will use descriptive statistics to summarize these data and examine correlations between NP characteristics and their effectiveness in promoting PrEP utilization. Qualitative feedback from NPs on the chatbot’s functionality, its integration into their workflow, and its impact on patient engagement will be transcribed and analyzed using the same 5-step framework analysis approach mentioned earlier. This analysis will help us identify specific barriers and facilitators NPs face in utilizing the chatbot for PrEP care and how the chatbot can be optimized to support their efforts. By combining these methods, we aim to provide a comprehensive understanding of the chatbot’s usability, effectiveness, and potential areas for improvement, particularly in facilitating PrEP utilization among Black cisgender women and supporting NPs in their role.

Upon submitting this manuscript, we obtained IRB approval for the research activities of recruiting participants to gather their preferences of chatbot features (University of Rochester Research Subjects Review Board: STUDY00009105).

All participants will provide informed consent before participating in the study. The consent process will be conducted on the web via REDCap, a secure web-based platform designed for research data collection and management. An information sheet detailing the study’s purpose, procedures, risks, and benefits will be provided to participants prior to study participation. Consent will be documented through an electronic consent form that participants will complete before any study activities begin.

Data collected during the study will be anonymized and deidentified to protect participants’ privacy. All data will be stored securely on the university’s servers, and access will be restricted to authorized study team members only. Measures such as encryption, secure servers, and compliance with HIPAA regulations will be implemented to safeguard participant information. Audio recordings of interviews will be transcribed and deidentified before analysis, and all identifiers will be deleted once transcription is complete.

Participants will be compensated for their time and contributions to the study. Black cisgender women participating in the in-depth interviews will receive US $30 each, while CAB members will receive US $20 for each meeting, for a total of 3 times throughout the study. Additionally, a small incentive of US $5 will be provided for each successful referral made through snowball sampling, with a cap on referral payments at 3 successful enrollments. Compensation will be provided irrespective of the completion status of the interviews or activities to ensure fairness.

Our proposed study harnesses AI technology and nursing skills through an AI-powered chatbot designed to facilitate PrEP care utilization among Black cisgender women. This culturally responsive, nurse-led, and AI-powered chatbot encapsulates biomedical, psychological, and sociocultural factors affecting PrEP utilization, making it a holistic tool for HIV prevention. Our model is scalable both geographically and technologically, making it capable of addressing health disparities beyond the immediate context of this study. As AI improves with increased data and user engagement, we anticipate the AI chatbot will significantly improve PrEP awareness, adherence, and retention among Black cisgender women by providing personalized, accessible, and stigma-free support.

Implementing AI-powered chatbots in health care has shown promise in various domains, including mental health support, chronic disease management, and patient education [13-15]. Previous studies have demonstrated that digital technology-assisted health care can improve patient engagement and outcomes [13-15]. However, few studies have specifically focused on using AI chatbots to address HIV prevention among Black cisgender women [11,12]. Our study builds on existing research by integrating cultural competence and stakeholder feedback into the chatbot design, making it uniquely tailored to the needs of this population. Our approach offers a novel solution by combining AI technology with nursing expertise, bridging the gap between digital health innovations and culturally sensitive care.

Implementing AI-powered chatbots presents several challenges and strengths [24,37,38]. Expected limitations include the following. First, technology accessibility: PrEP-eligible individuals without consistent access to technology or who are less comfortable with digital tools may face unintended exclusion, potentially reinforcing existing health care disparities. Second, data privacy: ethical considerations regarding data privacy and security are critical, as chatbots gather sensitive health information. Ensuring HIPAA compliance, robust encryption, and clear communication about data use is essential to maintain users’ trust. Third, generalizability: while our study focuses on Black cisgender women, the findings may not be generalizable to other populations without further adaptation and testing. On the other hand, there are several strengths identified in the proposed study. First, cultural competence: our chatbot is designed to meet the unique needs of Black cisgender women by incorporating their cultural contexts and addressing their specific concerns, enhancing its relevance and effectiveness. Second, scalability: the model is scalable both geographically and technologically, making it capable of addressing health disparities beyond the immediate context of this study. Third, holistic approach: the integration of biomedical, psychological, and sociocultural factors provides a comprehensive approach to HIV prevention, potentially leading to better health outcomes.

As the AI component evolves with more data and user interaction, we foresee opportunities for further technological advancements, enhancing the efficacy of our model. Upon completing this pilot, we aim to roll out this innovative program to other marginalized populations to expand its reach and impact using a subsequent large-scale hybrid type I study design. By doing so, we aim to contribute tangible, scalable solutions to the pressing challenge of racial disparities in HIV prevention. This vision aligns seamlessly with the funding foundation’s key priorities, including challenging conventional strategies, demonstrating potential as a best-in-class intervention, narrowing gaps in health equity, and showcasing the potential for scalability. We are deeply committed to realizing this vision and driving forward the fight against racial disparities in HIV prevention and beyond.