中医临床辅助诊疗技术发展与展望:从大语言模型到智能体OA

Traditional Chinese medicine(TCM)clinical decision support has recently shown a clear trend of technological transition.Earlier studies were mainly built on conventional deep learning frameworks,whereas current research is increasingly influenced by large language models(LLM).More recently,the emergence of agent-based methods has opened up new possibilities for interactive and collaborative diagnostic support.In this context,this review re-examined the development of TCM-oriented intelligent diagnosis and treatment,with particular attention to recent methodological changes,unresolved problems,and possible future direc-tions.On this basis,the review further turns to recent progress in medical agents.At the same time,research on TCM-specific agents is still at an early stage.Important issues,such as multi-turn closed-loop interaction,simulation of collaborative clinical workflows,and evaluation of reasoning interpretability,have not yet been adequately studied.Addressing these questions will be important for moving beyond proof-of-concept systems toward more practical and trustworthy applications.Overall,this review aimed at providing a clearer under-standing of how LLM and agents may reshape TCM clinical decision support in the coming years.