Application of Large Language Models in Traditional Chinese Medicine: A State-of-the-Art Review

Large language models (LLMs) are reshaping the landscape of Traditional Chinese Medicine (TCM). This review covers the latest applications of LLMs in TCM, including literature analysis, data mining, TCM knowledge management, diagnosis simulation and clinical decision making. LLMs can analyze large quantities of TCM literature and medical records to extract critical information, classify prescriptions, and build TCM knowledge maps to help researchers quickly grasp state-of-the-art and future research trends. LLMs can provide initial diagnostic recommendations by analyzing textual information such as a patient’s symptom description and medical history, enabling the optimization of TCM therapy and the training of TCM practitioners. Compared with traditional tools, LLMs can significantly improve the efficiency and accuracy of bibliographic analysis and TCM prescription classification, and offer new potential for data-driven standardized TCM diagnosis. However, challenges remain, including the standardization of TCM terminology and data formats, integration of different data sources, timely knowledge updates, and the interpretability and credibility of results generated by LLMs. Future research on standardized templates for patient symptom description, multimodal data fusion techniques, and real-time knowledge update systems is warranted to improve the transparency and interpretability of LLMs. This review highlights the potential of LLMs to modernize TCM research and practice, providing an up-to-date reference for data scientists, biomedical engineers, and TCM practitioners.