Parental Perspectives on Using Large Language Model-Powered Chatbots in Rare Disease Care

Parents of children with rare pediatric illnesses have high information needs. Large language model chatbots are AI tools parents can use to seek information. Parents are generally open to incorporating novel technologies into their child's medical care, yet more research is needed to understand their perceptions of using chatbots and their broader implementation in pediatric care.

We implemented a secure Health Insurance Portability and Accountability Act-compliant chat interface backed by a generative pretrained transformer (GPT-4o) and conducted think-aloud sessions with parents of children with cancer (n=24) and parents of children with vascular anomalies (n=24). We observed their use and explored their perceptions of the tool. We analyzed the interview transcripts using thematic analysis.

Parents felt the chatbot could help them accomplish tasks such as seeking reassurance, managing uncertainty, preparing for visits, supplementing clinician information, adapting to new developmental stages, coping with long wait times, disseminating information, and feeling empowered. They expressed a desire for tools that rapidly generate credible information synthesized across multiple sources in a format that is well organized and archived for future use. They also perceived several risks and ethical concerns regarding chatbots, including the potential for inaccurate or overwhelming information and overreliance on the chatbot over expert medical advice.

Our findings demonstrate that parents perceive chatbots to be a useful tool in advocating for care and reducing caregiver burden. However, their perceptions regarding trust, overreliance on AI, and the impact of AI on their relationship with their child's doctors underscore the need for more research on the use of AI chatbots in pediatric medicine. (Funded by the Siteman Cancer Center and Siteman Kids at St. Louis Children's Hospital.).