Examining Potential Usability and Health Beliefs Among Young Adults Using a Conversational Agent for HPV Vaccine Counseling

"...the potential for conversational agents to be an impactful tool for health promotion endeavors..."

When recommending vaccines to patients, physicians who practice patient-centred communication may be able to address perceptions and concerns about vaccination in ways that promote shared understanding, trust, and, possibly, increased vaccine uptake. Yet factors including inadequate communication training for physicians and time constraints contribute to the fact that, according to some studies, one-third to one-half of eligible patients do not receive any recommendation for the human papillomavirus (HPV) vaccine from their provider. This study explores the possibility of automating the HPV vaccine counseling experience in a way that could inspire patients - in this case, young college adults, who may have missed the HPV vaccine during their adolescence - to initiate discussion about the HPV vaccine when they see their provider.

Speech-based dialogue systems are software ("conversational") agents that carry discourse with a user involving several speech turns. In their previous work (see Related Summaries, below), the researchers performed an initial usability trial on using a conversational agent for HPV vaccine counseling among parents with at least one child under 18. The trial was conducted using the Wizard of OZ protocol, which simulates natural language interface tools for agents or robots with a human operator ("wizard") in a remote area that is directing dialogue interaction of the machine. The overall takeaway was receptiveness towards the ease of use and features of the agent, including interactivity of the dialogue and real-time question answering, useful information, and clarity of communication.

Conducted in April 2019, this study applied the same procedure on a sample of 24 college-aged adults at a public university in Texas, the United States (US). HPV vaccination rates in Texas are much lower than in the rest of the US (32.9-26.6% for boys and 39.7% for girls). The conversational agent, "Beverly", initiated a conversation with the user and about HPV and the HPV vaccine for 20-30 minutes. The drone operator utilised a script designed by study personnel to counsel on the HPV vaccine. After the simulated counseling information session, each participant completed a usability survey involving the System Usability Scale (SUS) and Speech User Interface Service Quality (SUISQ), as well as a health belief model survey, the Carolina HPV Immunization Attitudes and Beliefs Scale (CHIAS).

Summary of results:

• The overall usability (i.e., satisfaction, effectiveness, and efficiency), measured through the SUS score indicated that participants perceived the agent to have high usability that is slightly better or equivalent to other voice interactive interfaces. In addition, individuals who never received the HPV vaccine rated the system high at 80, which indicates that this tool could be particularly effective for individual users who had never received the HPV vaccine and could perhaps inform and encourage them better than other paper-based methods.
• Among the individuals who never received the HPV vaccine, there was a strong inverse relationship between SUISQ Overall Score and one health belief model construct: perceived effectiveness of the HPV vaccine. This would indicate that unvaccinated participants who liked the conversational agent are more likely to believe that the HPV vaccine is effective and consider getting vaccinated. There was also a marginally significant statistical difference with the perceived barriers construct.

According to the researchers, this study "provides initial evidence on the validity of using a conversational agent to improve vaccination rates. While these results are preliminary at best and lack a large sample size, there is some promise that deploying this type of tool within a health care environment could be feasible, especially since conversational agents promote patient-centered communication, which is important to achieve better health outcomes. Also, the relatively good usability rating which encompasses overall satisfaction could be analogous to patient satisfaction, which is an important outcome of patient-centered communication....In addition, the implementation of conversational agents in a health setting could possibly address some of the communication challenges rooted in demographic disparities (language barriers, low health literacy, etc.), and offload some of the communication burdens that impede on providers' time....This work could potentially be extended to other vaccine needs and health topics as well."