大语言模型时代中医临床决策支持的方法论转向与认知建模研究OA

Research on TCM clinical decision support is an important embodiment of keeping pace with the times.From a methodological perspective,this article systematically reviewed the evolution,major approaches,and core methodological challenges of TCM clinical decision support research,and argued that the deeper challenges arise from a misalignment between the cognitive characteristics of TCM and prevailing artificial intelligence modeling paradigms.In response to the new opportunities presented by the era of large language models(LLMs),this article proposed leveraging the strengths of LLMs in natural language understanding,multi-source knowledge integration,and complex semantic generation to facilitate a methodological shift in TCM clinical decision support—from task-oriented modeling to cognitive modeling.Within this framework,the plurality and complexity of TCM cognition can be effectively represented.Through knowledge constraints,structured reasoning chains,and feedback mechanisms,cognitive modeling and alignment in TCM can be achieved,enabling LLMs to be organically integrated into the clinical cognitive system.The methodological framework of cognitive modeling can not only lay the foundation for high-level modeling while preserving the disciplinary complexity and intrinsic characteristics of TCM,but also provide theoretical references for research on TCM-oriented artificial intelligence.