基于方剂相似度的核心方药及其适应症挖掘方法研究——以失眠症为例

Prescription Similarity-Based Analysis of Core Formulas and Medicinals and Related Indications through Data Mining:Taking Insomnia as an Example

目的探索建立中医师辨证论治核心方药的数据挖掘方法。方法以中医师辨证治疗失眠症的临床诊疗数据为例,基于方剂相似度对目标人群进行模块划分,构建核心方药及其有效人群特征的复杂网络,采用子网筛选、分级核心子网挖掘算法,挖掘中医师辨证治疗失眠症有效人群的核心方药及其适应症。结果共采集101例失眠症患者245诊次诊疗数据,其中有78例就诊1次以上,主要疗效结局显示有效患者68例(87.2%),部分有效患者2例,无效患者8例。78例(144诊次)患者诊疗数据进行复杂网络挖掘,根据各诊次方剂相似度可划分为2个模块,模块度为0.184。其中,模块一核心方由枳实、半夏、青礞石、竹茹、茯苓、黄连、橘红、大黄、龙齿、珍珠母组成,其有效人群以舌红、苔黄腻、入睡困难、早醒、多梦为主要表现,兼见大便干、口干口苦、情志不宁、头晕、脉弦实等。结论基于方剂相似度的复杂网络挖掘方法,能够初步揭示中医师辨证论治用药特点,为辅助名老中医经验挖掘和传承提供方法。

Objective To explore the data mining method for establishing the core formulas and medicinal herbs for traditional Chinese medicine(TCM)doctors to give treatment based on syndrome differentiation.Methods By taking insomnia as an example,the data mining method was explored to analyze the clinical data in TCM practice.Based on the similarity of prescriptions,the target population was identified through classification models;the complex network of core formulas and medicinals,and the characteristics of people benefited from the prescription was constructed;subnetwork screening as well as a hierarchical core subnetwork mining algorithm were adopted to explore the core formulas and medicinals for insomnia patients who were benefited from the TCM treatment.Results Data from 101 insomnia patients involving 245 hospital visits were collected,among which 78 patients visited the hospital more than once.Sixty-eight patients(87.2%)showed effective result,while two as partially effective,and eight as non-effective.A complex network of clinical data of 78 patients(144 visits)were constructed for data mining.Target population was divided into two modules according to the similarity of prescriptions,and the module degree was 0.184.The core formula of module one was composed of Zhishi(Fructus Aurantii Immaturus),Banxia(Rhizoma Pinelliae),Qingmengshi(Lapis Chloriti),Zhuru(Caulis Bambusae in Taenia),Fuling(Poria),Huanglian(Rhizoma Coptidis),Juhong(Exocarpium Citri Rubrum),Dahuang(Radix et Rhizoma Rhei),Longchi(Dens Draconis),and Zhenzhumu(Concha Margaritiferae Usta);it was effective for patients who mainly had a red tongue,and yellow and greasy coating;had difficulty falling asleep;waked up early;and had many dreams;certain patients may present dry stool,dry mouth,bitter taste,uneasy feelings,dizziness,wiry and excess pulse.Conclusion The prescription similarity-based complex network data mining method can be used to initially reveal the characteristics of using drugs based on syndrome differentiation by TCM doctors,the result of which may further facilitate the data mining and inheritance of TCM famous doctors′experience.