摘要
目的通过构建数据挖掘专用模型,探究中药方的君臣佐使配伍规律。方法针对中药方蕴藏的君臣佐使隐性配伍结构,结合中医君臣佐使理论内涵,构建一种基于贝叶斯网络理论的双关主题模型(DCTM),提出并采用基于DCTM的君臣佐使配伍规律分析方法。以中国方剂数据库共84464首中药方作为分析对象,以《伤寒明理药方论》标注的20首中药方的君臣佐使为参照标准,对DCTM分析结果进行评价。结果采用基于DCTM的君臣佐使配伍规律分析方法对20首中药方标注结果的君药、臣药、佐药、使药准确率分别为0.75、0.50、0.71、0.81,召回率分别为0.75、0.45、0.85、0.43,F1-得分分别为0.75、0.47、0.77、0.56。结论本研究提出的挖掘君臣佐使隐性药方结构的DCTM模型和方法能够有效分析出君臣佐使角色,可为印证君臣佐使理论、完善中医君臣佐使配伍方法提供帮助。
Objective To explore the compatibility rules of the chief,deputy,assistant and envoy in TCM prescriptions by constructing a special model for data mining.Methods Aiming at the hidden compatibility structure of chief,deputy,assistant and envoy contained in TCM prescriptions,combined with its theoretical connotation,a double correlated topic model(DCTM)based on Bayesian network theory was constructed.Furthermore,an analysis method for compatibility rules based on DCTM was proposed.A total of 84464 prescriptions in China Prescription Database were used as the analysis objects.20 prescriptions marked in Shang Han Ming Li Yao Fang Lun were set as the reference standard of chief,deputy,assistant and envoy.The analysis results were evaluated.Results The chief,deputy,assistant and envoy of 20 prescriptions were marked by the anlysis method based on DCTM.The precision rate of chief,deputy,assistant and envoy marked by this method was 0.75,0.50,0.71 and 0.81,respectively.The recall rate was 0.75,0.45,0.85,0.43,respectively.The F1 score reached 0.75,0.47,0.77,0.56,respectively.Conclusion The DCTM model and method proposed in this study can effectively analyze the relevant roles of chief,deputy,assistant and envoy,which can provide help for confirming the theory of chief,deputy,assistant and envoy,and improving the compatibility method of chief,deputy,assistant and envoy in TCM.
作者
马甲林
王兆军
郭海
MAJialin;WANG Zhaojun;GUO Hai(College of Computer Science,Huaiyin Institute of Technology,Huaian 223003,China;Huaiyin Wu Jutong Institute of Traditional Chinese Medicine,Huaian 223300,China;The Affiliated Huaian No.1 People’s Hospital of Nanjing Medical University,Huaian 223000,China)
出处
《中国中医药信息杂志》
CAS
CSCD
2022年第12期23-29,共7页
Chinese Journal of Information on Traditional Chinese Medicine
基金
国家自然科学基金(61602202)。
关键词
组方结构
君臣佐使
数据挖掘
双关主题模型
prescription structure
chief,deputy,assistant and envoy
data mining
double correlated topic model
