目的 设计一种中药社团发现算法。方法 将中医处方转化为矩阵数据,通过矩阵运算找出不同中药之间的关联度,以中药为节点、关联度为边,构造关联网络,采用分裂的层次聚类方法对关联网络进行社团划分,建立中药社团发现算法HCD。为了验证算法的有效性,将HCD与经典的社团发现算法GN进行比较,分别对970诊次结肠癌病案资料进行分析,比较分析结果差异。结果 HCD能够较好地划分中药社团,划分结果符合中医理论,划分效果优于GN算法。结论 将关联网络和层次聚类相结合构造的中药社团发现算法HCD能够有效划分中药社团,其可以广泛应用于中医临床数据挖掘中,为名老中医诊疗挖掘提供支撑。
Objective To design a Chinese medicine community detection algorithm. Methods Firstly, the TCM prescriptions were transformed into matrix data, and the correlation degree (DCR) between different herbs was found out through matrix operation. Secondly, by taking herb as the node and relevance degree as the edge, the association network is constructed. The split hierarchical clustering method is used to divide the association network into communities, and then the detection algorithm HCD is established. In order to verify the effectiveness of the algorithm, HCD is compared with the classic community detection algorithm GN. Both of them were used to analyze the data of 970 cases of colon cancer respectively, and the results were compared. Results HCD can better divide the communities of traditional Chinese medicine and the result is in line with the theory of traditional Chinese medicine, the dividing effect is better than GN algorithm. Conclusion the detection algorithm HCD constructed by combining association network and hierarchical clustering can effectively divide TCM communities, which can be widely applied in TCM clinical data mining, providing support for the diagnosis and treatment mining of famous and old TCM.