Purpose/Significance The paper introduces the application status and challenges of generative language model in the medical field, and proposes a knowledge-enhanced medical language model to improve the specialization, accuracy and credibility of the model, and provides references for researchers in the fields of medicine, language model and knowledge graph. Method/Process It reviews the development, current status, and major technologies of large language models, and analyzes the challenges in data security, professionalism, ethics, and model interpretability. It introduces the common application scenarios and technical points of medical generative language model, and focuses on the medical language model based on knowledge graph and multi-modal data fusion knowledge enhancement, including the advantages, technical principles and specific cases. Result/Conclusion The knowledge-enhanced medical language model can improve the understanding, cognition and application capability of language model to professional medical knowledge, enhance the generative capability of natural language, and expand the processing capability of multi-modal data, which has a wide application prospect in medical question answering, intelligent assisted diagnosis, personalized medical decision making and so on.