Abstract: The practice of evidence-based medicine requires clinicians to consider and integrate patients' values and preference (V&P) in clinical decision-making to achieve personalized diagnosis and treatment. However, traditional methods have many limitations in accurately obtaining and integrating patients' V&P. Large language models (LLM), with their exceptional natural language processing capabilities, have brought new opportunities and transformative changes to the medical field. Large language modeling (LLM), with exceptional natural language processing capabilities, have introduced new opportunities and transformations in the medical field. LLM demonstrates substantial potential in assisting with patient V&P research, predicting patient preferences, and simulating decisional conflict, thereby significantly enhancing the efficiency and quality of data collection and analysis. This paper summarizes the relevant research progress of LLM in the medical field at home and abroad and deeply explores its prospects and advantages in the study of patients' V&P. It aims to provide guidance for clinicians and researchers, promote the rational application of LLM in V&P research, assist in the development of evidence-based medicine, and ultimately achieve the patient-centered medical model.