基于数据挖掘的高血压性眩晕医案辨证用药特点分析

Data Mining-Based Analysis of Medication Characteristics of Syndrome Differentiation in Medical Cases of Hypertensive Vertigo

目的:基于数据挖掘技术研究高血压性眩晕组方用药规律,归纳辨治特色,指导临床运用。方法:通过检索各数据库中医家医案文献,构建方药数据库,通过SPSS Modeler 18.0、SPSS Statistics 22.0、中医传承辅助系统平台(V2.5)对组方用药进行频次统计、关联规则、复杂网络、系统聚类、熵层次聚类新方及主成分因子分析。结果:253首处方涉及中药258味,涵盖7个病位证素和13个病性证素;挖掘出用药频次大于平均频次的药物61味,高频中药是天麻、钩藤、甘草、茯苓、白芍、牛膝,病位证素以肝、脾、肾为主,病性证素以阳亢、阴虚、痰、动风、血瘀、湿为主;当最小支持度为10%,最小置信度为75%,得到药对关联组合19项,当最小支持度为10%,最小置信度为90%,得到三项关联组合20项;复杂网络生成核心组方为天麻、钩藤、半夏、茯苓、牛膝、白术、石决明、甘草;系统聚类得到龙骨、牡蛎等16组药物聚类组合,熵层次聚类得到10个中药新处方,主成分因子分析共提取出22个公因子。结论:高血压性眩晕多属本虚标实之证,病机多见阴虚阳亢动风,治疗以滋阴平肝、熄风止眩为基本法则,随证搭配。

Objective:To study the medication rules for hypertensive vertigo and summarize the characteristics of syndrome differentiation and treatment based on data mining technology,in order to better guide the clinical application.Methods:The medical cases written by doctors of traditional Chinese medicine(TCM)were retrieved from each database for constructing a database of related prescriptions and Chinese herbs.The frequency statistics of prescriptions and Chinese herbs,association rule analysis,complex network analysis,systematic clustering,entropy hierarchical clustering of new prescriptions,and principal component factor analysis were carried out using SPSS Modeler 18.0,SPSS Statistics 22.0 and Traditional Chinese Medicine Inheritance Support System(V2.5).Results:A total of 253 prescriptions involving 258 Chinese herbs were screened out,covering seven disease locations and 13 disease natures.There were 61 Chinese herbs exhibiting the frequency of usage higher than the average.The high-frequency Chinese herbs were Tianma(天麻),Gouteng(钩藤),Gancao(甘草),Fuling(茯苓),Baishao(白芍),and Niuxi(牛膝).The disease was mainly located in Gan(肝),Pi(脾),and Shen(肾).In terms of the disease nature,it could be mainly differentiated into Yangkang(阳亢),Yinxu(阴虚),Tan(痰),Dongfeng(动风),Xueyu(血瘀),and Shi(湿).When the minimum support degree was 10%and the minimum confidence coefficient was 75%,19 two-item association combinations were obtained.When the minimum support degree was 10%and the minimum confidence coefficient was 90%,20 three-item association combinations were harvested.The core prescription was generated by complex network analysis consisted of Tianma(天麻),Gouteng(钩藤),Banxia(半夏),Fuling(茯苓),Niuxi(牛膝),Baizhu(白术),Shijueming(石决明),and Gancao(甘草).16 Chinese herbal clusters such as Longgu(龙骨)-Muli(牡蛎)were yielded by the systematic clustering,ten new prescriptions were obtained by the entropy hierarchical,and 22 common factors were analyzed by the principal component factor analysis.Conclusion:Hypertensive vertigo is mostly characterized by Benxu(本虚)and Biaoshi(标实)and results from Yinxu(阴虚),Yangkang(阳亢),and Dongfeng(动风).The basic therapeutic principles lie in Ziyin(滋阴),Pinggan(平肝),Xifeng(熄风),and Zhixuan(止眩),which can be combined according to the actual syndrome.