目的 以名中医治疗甲状腺结节的临床病案为研究对象，探索运用不完备信息系统下混合增量式约简发现诊疗规律，为传承名老中医学术思想、精准化临床用药提供参考。方法 基于国医大师周仲瑛教授及其团队的临床甲状腺结节诊疗医案，构建混合增量式信息系统，以适应数据集不断变化的临床医案要求，研究混合增量式学习，提出对象增加时的增量式区分矩阵的属性约简算法，分析核心药物、治法、病机和证型。结果 核心病位以肝为主，多涉及脾、肾、痰、瘀、郁、火、虚为其主要病理因素。证型分分为气郁型、痰湿型、湿热型、血瘀型、及阳虚型。结论 在临床医案数据上运用增量式信息系统，构建基于增量式区分矩阵，可有效的提取核心方、核心药，探索“法-方-药”之间的联系，为传承名老中医学术思想，指导临床用药提供参考。

Objective Taking the clinical medical records of famous traditional Chinese medicine in the treatment of thyroid nodules as the research object, to explore the application of mixed incremental reduction under incomplete information system to discover the diagnosis and treatment rules, and to provide reference for inheriting the academic ideas of famous traditional Chinese medicine and precise clinical medication.Methods Based on the clinical medical records of thyroid nodules by Professor Zhou Zhongying, Chinese medicine master, and his team, a hybrid incremental information system was constructed to adapt to the changing requirements of clinical medical records. The hybrid incremental learning was studied. The attribute reduction algorithm of incremental discernibility matrix was proposed when the objects increased. The core drugs, therapy, pathogenesis and syndrome types were analyzed.Conclusion Based on the incremental discernibility matrix, the incremental information system can effectively extract the core prescription and core medicine, and explore the relationship in "method-prescription-medicine", which can provide reference for inheriting the academic thoughts of famous and experienced TCM doctors and guiding clinical medication.

R249