fbpx
Health and Medicine

Chatbot informs women about HIV prevention

Developed by Carolina researchers, ShesPrEPared uses artificial intelligence to help them pick their best option.

Animated image of phone receiving text notifications.
(Zack Hall/UNC-Chapel Hill)

What if ChatGPT attended medical school, then developed specialized expertise through a residency in HIV prevention for women?

That’s how its creators describe ShesPrEPared, an artificial intelligence chatbot and website being developed by an interdisciplinary team at UNC-Chapel Hill. The new chatbot will help women worldwide learn about and choose their preferred HIV prevention method.

Leading the project is Lauren Hill, an assistant professor in the UNC Gillings School of Global Public Health’s health behavior department. Her research centers on community- and clinic-based interventions to promote women’s understanding and effective use of PrEP, which stands for preexposure prophylaxis. PrEP is taken for protection before exposure to HIV, the virus that causes AIDS.

Hill’s collaborators include Shashank Srivastava and Dr. Christopher Sellers. Srivastava is an assistant professor in the UNC College of Arts and Sciences’ computer science department who is an expert in natural language processing, AI, machine learning and their applications. Sellers is an assistant professor in the UNC School of Medicine and medical director of Wake County’s Human Services HIV and sexually transmitted infection clinics.

The team aims to create a user-friendly website and a chatbot that uses a shared decision-making approach, a best medical practice. Users will be able to ask any question about PrEP. The answers will include specialized information that other chatbots cannot provide and tailored support to help users decide about their preference for HIV prevention.

Srivastava’s team is building the ShesPrEPared chatbot on a ChatGPT foundation but with a deeper knowledge. “It’s like ChatGPT went to college and knows some stuff about HIV and PrEP, but we’re training our chatbot like it’s completing a medical residency or specialist training. It knows more about PrEP that is specific to women, current and accurate,” Hill said.

Hill adopts a “a real-world impact” view of AIDS prevention, focusing on women’s health from adolescence to adulthood. “Some periods in life are potentially more important than others in determining one’s health. Pregnancy and the postpartum period, for a lot of women, particularly in Africa, are when they are most vulnerable to HIV,” she said.

The chatbot is part of a process that begins with education about HIV prevention. “In public health, we know that information isn’t enough,” Hill said. “As a behavioral scientist, I try to help people take actions that they want to take for their health, being realistic about their life circumstances, their resources and whatever may either help or hinder them to live a healthy life.”

HIV-treatment drugs have existed since 1987, when AZT gained FDA approval. HIV-prevention drugs have only been in use since 2012. Three HIV-prevention drugs – two in pill form and one monthly injection – and a vaginal ring are approved for use.

“Just because these great products exist doesn’t mean that people will use them or use them effectively. Product selection needs to center on the person’s preferences,” Hill said. The team wants people to use the website and chatbot to learn about the products’ pros and cons, so that they can determine which products are best for them.

North Carolina women considering HIV prevention will be recruited to test the chatbot through local clinics. The long-term goal is to make the chatbot available in waiting rooms and as part of medical provider referral so women can learn about PrEP before meeting with a provider.

The Gillings Innovation Lab and the UNC Center for AIDS Research are funding the team’s work.