Balancing Health Information-Seeking through Retrieval-Augmented Generation-Based LLM Chatbot

Family caregivers play a vital role in supporting children with chronic health conditions, such as neonates diagnosed with hypoxic-ischemic encephalopathy (HIE). However, navigating complex medical information can be overwhelming due to the quantity and quality of available literature. This study leverages Retrieval-Augmented Generation (RAG)-based Large Language Models (LLMs) to develop a chatbot that integrates peer-reviewed scientific literature and provides personalized, simplified summaries for caregivers. A user study involving six caregivers and five healthcare providers demonstrated the chatbot’s ability to enhance clarity, improve comprehension, and deliver essential medical information concisely. Our findings highlight the potential of RAG-based LLMs to enhance caregivers’ health literacy and support their information-seeking behavior, while also underscoring the importance of thoughtfully navigating the differing expectations of caregivers and healthcare providers regarding the type, depth, and presentation of medical information.