According to the results of the National Nutrition and Health Survey conducted by the Health Promotion Administration (HPA), Ministry of Health and Welfare, the overall level of nutritional literacy and health education knowledge among the Taiwanese population remains insufficient. Furthermore, the top ten leading causes of death in 2023 were closely linked to lifestyle factors, underscoring the critical need for improved health awareness to enhance health outcomes and quality of life. Consequently, providing health recommendations tailored to individual needs has become a vital issue in contemporary health management. Against this background, this study develops a personalized health recommendation model that integrates Personal Health Records (PHRs) and leverages Large Language Model (LLM) technologies to construct an intelligent system capable of delivering health consultations and recommendations.
Traditional medical models are predominantly disease-oriented and rely on standardized treatments, which often fail to adequately reflect individual differences in lifestyle habits and health conditions. Meanwhile, although LLMs have increasingly been used
to generate personalized health recommendations, their outputs remain constrained by the scope of the training data and may lack sufficient completeness or accuracy when addressing specialized medical issues. To address these limitations, this study adopts a Retrieval-Augmented Generation (RAG) framework to construct a health information database that integrates user queries, PHR data, and external domain knowledge. The results demonstrate that the proposed framework can generate knowledge-based, practical health recommendations tailored to individual health conditions. These findings not only enhance users’ health literacy but also provide a significant reference for integrating generative AI into future healthcare management.