基于决策树的《叶氏女科证治》常用中药用药规律挖掘研究

Study on Medication Law of Commonly Used TCM Chinese Materia Medica in Ye Shi Nyu Ke Zheng Zhi Based on Decision Tree

目的通过数据挖掘中的决策树C5.0模型,探讨《叶氏女科证治》常用中药的用药规律。方法选择《叶氏女科证治》中调经、安胎、保产及求嗣中妇人相关内容,建立方药数据库,采用SPSS Modeler 18.0软件对高频中药使用C 5.0模型生成决策树及规则集。结果对频次≥200的当归、白芍、川芎、人参、白术、茯苓和熟地黄进行决策树分析,所得模型树深度均≤4,提示规则均具有普遍性;准确率较高,可控制在≥75%水平;模型受试者工作特征曲线下面积(AUC)值均在0.70~0.85,模型效果一般,但组成四物汤的当归、白芍、川芎和熟地黄,AUC值均≥0.8,模型效果相对较好。结论决策树模型可更清晰展示《叶氏女科证治》用药规律,有助于理解其用药思路,亦为其常用药提供数据支持。

Objective To explore the medication law of commonly used Chinese materia medica in the book of Ye Shi Nyu Ke Zheng Zhi through the decision tree C 5.0 model in data mining.Methods The relevant contents of women in regulating menstruation,preventing miscarriage,ensuring labor and seeking pregnancy in the book of Ye Shi Nyu Ke Zheng Zhi were selected to establish a prescription database.SPSS Modeler 18.0 software was used to generate a decision tree and rule set for high-frequency Chinese materia medica using C 5.0 model.Results The decision tree analysis was carried out on the frequency of≥200 Angelicae Sinensis Radix,Paeoniae Radix Alba,Chuanxiong Rhizoma,Ginseng Radix et Rhizoma,Atractylodis Macrocephalae Rhizoma,Poria and Rehmanniae Radix Praeparata.The depth of the model tree was≤4,suggesting that the rules were universal;the accuracy rate was high,which could be controlled at the level of≥75%;the AUC values of the model were 0.70-0.85,and the model effect was general.However,the AUC values of Angelicae Sinensis Radix,Paeoniae Radix Alba,Chuanxiong Rhizoma and Rehmanniae Radix Praeparata in Siwu Decoction were all≥0.8,and the model effect was relatively good.Conclusion In the form of decision tree model,the medication law of Ye Shi Nyu Ke Zheng Zhi can be more clearly displayed,which is helpful to understand its medication ideas and provide data support for its commonly used drugs.