复杂系统熵网络方法及其在中医肿瘤临床数据挖掘中的应用  

Complex Systems Entropy Network and Its Application in Data Mining for Chinese Medicine Tumor Clinics

作　　者：杨铭[1] 焦丽静[2] 陈佩奇[2] 王珏[2] 许玲[2] 

机构地区：[1]上海中医药大学附属龙华医院药剂科,上海200032 [2]上海中医药大学附属龙华医院肿瘤科,上海200032

出　　处：《世界科学技术-中医药现代化》2012年第2期1376-1383,共8页Modernization of Traditional Chinese Medicine and Materia Medica-World Science and Technology

基　　金：恶性肿瘤国家中医临床研究基地;国家中医药管理局中医肿瘤病学重点学科;上海市卫生局科研基金项目(2010Y121):基于复杂系统的中医肿瘤方剂网络的构建;负责人:杨铭;上海市医院药学科研基金项目(2010-YY-04):基于复杂网络的COPD中医证型及其用药相关性的研究;负责人:杨铭

摘　　要：目的:探索中医治疗肿瘤复杂临床数据挖掘的研究方法。方法:采用基于随机置换检验的互信息方法进行多维的相关性评估,在此基础上建立复杂网络,通过计算网络的各种统计信息对复杂中医肿瘤临床数据进行挖掘,并对龙华医院718例肺癌数据进行了挖掘分析。结果:中医治疗肿瘤临床数据中的多因素混杂交互效应,对718例数据分析结果得到116对有协同效应的属性,82对有拮抗效应的属性,14个核心协同效应子群属性集,7个核心拮抗效应子群属性集,分析结果基本符合临床实际情况。结论:复杂系统熵网络方法能够对复杂中医临床数据进行挖掘,是进行多因素交互效应分析的有效方法之一。This study was aimed to investigate the method of data mining for complex Chinese medicine tumor clinics data. This paper puts forth complex systems entropy network for data mining in tumor clinics. A mutual information al- gorithm based random permutation test is proposed to assess the correlation of multi-variables and then complex net- work is established. Based on tumor clinical data （718 cases） collected, data mining was performed by statistical infor- mation of the complex network. The results showed that interaction effects among multi-variables were discovered by complex systems entropy network. A total of 116 pairs synergy variables and lg core synergy cliques, 82 pairs antago- nistic variables and 7 core antagonistic cliques were found after 718 cases data mining. The results, for the most part, correspond with the actual clinics. It was concluded that complex systems entropy network is suitable for analysis of interaction effects among multi-variables and data mining for complex Chinese medicine clinics.

关 键 词：互信息 随机置换检验 复杂网络 数据挖掘 中医肿瘤 

分 类 号：R73[医药卫生—肿瘤]