基于层次聚类的失眠处方用药分析

Analyzing insomnia prescription drugs with hierarchical clustering

为了探寻层次聚类在失眠处方用药分析上的应用情况,进而分析失眠处方的用药规律,收集并整理了《方剂大辞典》中主治失眠的处方。对单味药物的四气、五味、归经及功效等数据,根据单连接、全连接和平均连接这三种不同的相似性度量方法进行层次聚类分析并比较。结果显示,基于全连接的层次聚类分组最为合理,将性味归经和功效有极大相似度的药物聚为一类,其聚类结果符合一定的中医理论。层次聚类结果客观地反映了失眠处方中药物间的关联关系,间接体现了失眠用药的药物组合规律,为临床用药提供新的研究方法和思路。

In order to explore the application of hierarchical clustering in the analysis of insomnia prescription medication,and then analyze the medication rules of insomnia prescription,the prescriptions for insomnia in the“Dictionary of Prescriptions”were collected and compiled.The dataset of four natures,five flavors,meridian tropism and efficacy of single-drug was analyzed and compared by hierarchical clustering analysis according to three different similarity measures of single linkage,complete linkage and average linkage.Results show that,the classification by hierarchical clustering with the complete linkage is the most reasonable,the drugs with the great similarity of four natures,five flavors,meridian tropism and efficacy become a cluster,and the clustering results are consistent with certain TCM theory.Hierarchical clustering objectively reflects the relationship between drugs in insomnia prescriptions,indirectly reflects the drug combination rules of insomnia medication,and provides new research methods and ideas for clinical drug use.