Potential Role of Conversational Agents in Encouraging PrEP Uptake

Abstract

Approximately 1.2 million people are living with HIV, with many of them unaware of their infection. Pre-exposure prophylaxis (PrEP) is available to minimize transmission among those at high risk for infection, including men who have sex with men, people who inject drugs, and female sex workers. Despite its availability, there is low usage of PrEP. To address this problem, various digital tools have been examined in HIV research. Among those, conversational agents are still underused and their capacity warrants examination to reach at-risk populations. In this paper, we discuss the potential of conversational agents in increasing uptake of PrEP by addressing barriers experienced among those at high risk for infection. BackgroundHIV continues to be a public health crisis. Approximately 1.2 million people are living with HIV in the USA and 14% are unaware of their infection. 1 Approximately 40% of community transmission can be attributed to individuals who are unaware of their infection. 2 Currently, there is no known cure for HIV. However, it is possible to minimize transmission from a person living with HIV to their sexual partner through pre-exposure prophylaxis (PrEP). PrEP decreases the risk of transmission when taken as prescribed, with efficacy increasing as adherence increases. 3,4 In 2016, the number of individuals taking PrEP was more than 77,000, with more males taking PrEP than females. 5 However, despite its availability, prescriptions for PrEP remain low. 6 Lack of awareness of availability by both at-risk individuals and people living with HIV, provider bias and distrust of the healthcare system, 6 stigma, 6,7 and low perception of HIV risk create barriers to uptake of PrEP.

Chatbots to Improve Sexual and Reproductive Health: Realist Synthesis

“…The literature on chatbots for SRH suggests that people value the anonymous and nonjudgmental space that chatbots offer for SRH discussions, particularly in contexts where SRH is stigmatized, or perceived as stigmatized, and for groups that face or perceive SRH stigma [17,18,[27][28][29][30][31][32][33][34][35][36][37][38]. The literature that supports stigma as a barrier, across many contexts, and the evidence on chatbots as a response to this, included high-income countries (n=5) and low-income countries (n=7).…”

“…Adolescent girls in this context may also be less comfortable about openly discussing SRH and may hold higher levels of privacy concerns due to taboos concerning SRH for girls [37]. Studies CMOCs [17,18,[27][28][29][30][31][32][33][35][36][37][38][44][45][46][47][48] In contexts where conversations about SRH a generate stigma and embarrassment (C), people may engage with chatbots (O) because chatbots are nonjudgmental and anonymous (M) [35,37] In contexts where conversations about SRH are taboo (C), some populations may not engage with SRH chatbots (O), because discussing SRH even through an anonymous medium remains stigmatizing (M) [33,[49][50][51] Where chatbots assure users that their information will be kept anonymous and their privacy will be maintained (C), users may engage with the chatbot (O), because their concerns have been addressed (M) [37,38,52,53] Where users do not have access to a private digital device (C), users are not afforded anonymity with chatbot use (O), because they cannot assure that their interaction with the chatbot will not be seen by other users of the device (M) a SRH: sexual and reproductive health.…”

“…Incorrect answers could generate health risks where users act on inappropriate clinical advice or signposting [59,60]. Studies CMOCs [17,18,29,30,35,48,49,57,61] When chatbots provide access to accurate information in digestible form (C), chatbots may be preferred to search engines (O), as the chatbot can eliminate steps to search and filter web-based health information (M) [18,31,58,62] When the language cues used, make chatbots feel uncanny (not quite human), like replying too quickly, misunderstanding, or overly formal language (C), then users can disengage from connecting with the Chatbot (O), as humans are sensitive to language cues that do not "feel right" (M) [18,30,35,38,50,53,58] When chatbots interact with users by prompting further questions and checking in with them (C), users engage for longer with the chatbot (C), because interaction drives the "conversation" between the user and chatbot forward and feels more human (M) [28,30] Where chatbots repeat information, either during a single session over repeated sessions (C), users may engage with the information provided (O), because repetition reinforces understanding (M) [30,52,58] Where chatbots use language that validates users' feelings and needs (C), this may engage users in chatbot use (O), because the chatbot offers a feeling of being understood (M) [54,56,62,63] Where chatbots give complex information on SRH topics (C), users may be able to understand the information more easily (O), because the information is given in a dialogical structure that shares information in short segments of "chunks" (O)…”

“…In these situations, chatbots may provide a rapid and timely assessment (eg, is emergency contraception needed) and signposting to wider services. Studies CMOCs [18,31,35,38,48,58] When chatbots offer 24-hour access to SRH a information (C), then users may find this convenient (O), because access to SRH information is not constrained by service opening times (M) [17,18,29,30,35,49,57] When chatbots provide immediate access to accurate information (C), users may stay engaged (O), as some users use chatbots in times of immediate need that require a fast response (M) a SRH: sexual and reproductive health.…”

The role of an artificial intelligence model in antiretroviral therapy counselling and advice for people living with HIV

“…Chatbots may improve access to accurate health‐related information for people living with HIV despite financial constraints or the social stigma associated with attending sexual health clinics. AI chatbots have been shown to potentially improve knowledge about and uptake of HIV pre‐exposure prophylaxis (PrEP) [5, 6]. We sought to extend this concept further to examine the potential utility of ChatGPT in the context of antiretroviral therapy (ART) counselling for patients with HIV.…”