Abstract
Personal health question answering (PHQA) systems have the potential to revolutionize healthcare delivery by providing individuals with personalized and accurate health information. However, existing PHQA systems often struggle to handle the complexity and nuance of natural language queries. This dissertation introduces a novel framework that leverages knowledge graphs (KGs) to enhance the capabilities of PHQA systems. Our framework integrates KGs with advanced query algorithms and natural language response generation techniques to deliver precise, contextually relevant, and personalized health information. We demonstrate that our KG-enhanced PHQA system, Medicient, outperforms conventional large language models in terms of accuracy and relevance. Furthermore, we showcase the effectiveness of iterative KG enhancement in progressively refining the system's responses. Finally, we demonstrate the generalizability of our approach by applying it to a domain outside of healthcare. This dissertation paves the way for the development of next-generation PHQA systems that can empower individuals with the knowledge and confidence to actively participate in their healthcare decisions.