目的 建立以异质关联网络为基础的辨证规律挖掘方法。方法 从医案数据入手，以矩阵运算为基础，以联合度为评价指标，构建“症状-证素-证型”异质关联网络HAN（Heterogeneous Associated Network），探索“症状-证素-证型”两两元素之间的组合规则。为了验证算法的有效性，与经典的关联分析算法Apriori进行比较，分别对1164条肝癌医案的辨证规律进行分析，比较两种算法的提取结果。结果 HAN算法提取结果和计算效率优于Apriori算法，提取辨证规律符合中医理论和专家经验。结论 利用HAN算法能高效精准地挖掘医案中症状、证素、证型之间的潜在关系，可以为名老中医临床经验挖掘提供方法参考。
Objective: To establish a new syndrome differentiation law mining algorithm based on heterogeneous associated network. Method: Based on matrix operation of TCM medical records,, the heterogeneous associated network (HAN), which was evaluated by unite degree, was constructed between symptoms,syndrome elements and syndromes. To validate the effectiveness of HAN algorithm,1164 prescriptions of liver cancer in TCM medical records were analyzed by HAN and Apriori separately, and syndrome differentiation rules were extracted. By comparing the extracted rules, the performances of HAN and Apriori were evaluated . Result: HAN performed better than Apriori both in extracting accuracy and computing efficicy and the syndrome differentiation laws mined by HAN were based on TCM theories and expert experience. Conclusion: HAN algorithm can effectively and accurately find potential relations between symptoms,syndrome elements and syndrome types. The results can provide reference for syndrome differentiation law mining of famous eld TCM doctors.