- September 08, 2025
- By Jessica Weiss ’05
The human papillomavirus (HPV) vaccine is one of the few vaccines proven to prevent cancer—so why have only 61% of teens 13-17 and far lower percentages of younger children received the potentially life-saving shots?
A new project led by University of Maryland communication Professor Xiaoli Nan aims to change those discouraging statistics by harnessing the power of artificial intelligence to help parents make informed choices.
Backed by a $2.8 million grant from the National Institutes of Health’s National Cancer Institute, Nan and an interdisciplinary team of researchers are developing a personalized, AI-driven chatbot that tailors health information to parents’ specific concerns and communication styles.
“This vaccine is safe, effective and prevents several types of cancer, but many parents still hesitate,” Nan said. “We’re using AI to help meet them where they are and make sure they feel supported in the decision-making process.”
The project also includes Cheryl Knott and Min Qi Wang, professors of behavioral and community health in the School of Public Health, Philip Resnik, an MPower Professor with a joint appointment in the Department of Linguistics and the University of Maryland Institute for Advanced Computer Studies, and Clement Adebamowo and Shana Ntiri from the School of Medicine at the University of Maryland, Baltimore.
Recommended for adolescents as early as age 9, the HPV vaccine is widely endorsed by physicians as a safe and lasting defense against cancers of the anus, cervix, throat and other areas. But misinformation, stigma and distrust have slowed its widespread adoption, said Knott, who is also associate director of community outreach and engagement at the University of Maryland Marlene and Stuart Greenebaum Comprehensive Cancer Center in Baltimore.
Part of the stigma comes from the vaccine’s link to a sexually transmitted infection. Some parents hear that and wonder if it’s too soon, choosing to delay or avoid vaccination altogether.
“The literature shows that this vaccine is underutilized especially in certain groups, for example men, those with less education, and those with less access to health care,” Knott said.
While traditional communication tools—like brochures and websites—often fail to address concerns at an individual level, the research team’s AI-powered chatbot will use natural language processing to simulate human conversation. But it goes further, tailoring messages to a user’s specific beliefs, concerns and communication preferences.
Unlike scripted chatbots that only deliver canned replies—or generative AI models like ChatGPT, which can produce unreliable text—the team’s system draws on large language models in a more controlled way. The models help generate a broad set of potential responses, which are then curated and vetted by the team’s experts. During live use, the chatbot analyzes what a parent has said and selects the most appropriate reply from that trusted set.
Resnik describes this as a “middle ground” approach: not limited to rigid, prewritten scripts, but also not the “wild west” of open-ended text where “guardrails” have often proven ineffective. “You have curated possibilities,” he said, “and at the same time much more flexibility, because the system is analyzing what the person says and adapting in response.”
The first phase of the project focuses on refining the chatbot through a user-centered design process and gathering input from parents, health professionals and community members. Once refined, the chatbot will be tested in two randomized controlled trials. The first will be conducted online with a national sample of U.S. parents, comparing the chatbot’s impact to a standard CDC pamphlet. The second trial will take place in clinical settings in Baltimore, including at pediatricians’ offices.
Beyond this chatbot’s focus on HPV, the project has far-reaching implications for public health communication in the age of AI, the researchers said. Beyond its potential to drive vaccine uptake, they hope to demonstrate how AI can help public health systems respond more quickly, thoughtfully and inclusively—especially in times of crisis. In future pandemics or other fast-moving public health challenges, tools like the chatbot could offer a scalable, adaptable solution for reaching people with timely, personalized guidance, Nan said.
The project “really makes people think about what’s possible,” she said. “Can we do a better job with public health communication—with speed, scale and empathy? I think the answer is yes.”