基于隐结构模型的高胆固醇血症证候要素评价  

Reviewing syndrome factors of hypercholesterolemia based on latent structure model

作　　者：何伟[1,2] 郭树仁[3] 谢雁鸣[4] 刘平兰[3] 田峰[2] 张燕[3] 赵君贤[3] 刘和平[3] 

机构地区：[1]陕西中医学院基础医学院,陕西712046 [2]中国中医科学院博士后流动站 [3]北京北大维信生物科技有限公司 [4]中国中医科学院中医临床基础医学研究所

出　　处：《北京中医药大学学报》2013年第12期812-816,共5页Journal of Beijing University of Traditional Chinese Medicine

基　　金：重大新药创制科技重大专项"十一五"计划资助项目(No.2008ZX09202-007)

摘　　要：目的评价高胆固醇血症的主要中医证候要素特征。方法采用多维分层聚类的隐结构模型方法,构建419例高胆固醇血症的隐树结构图,分析隐类概率、条件概率、互信息及累积信息覆盖率等量化数据,以人工判读方法诠释高胆固醇血症的主要证候要素特征。结果以累积信息覆盖率达到95%为主要证候要素的判定标准,419例高胆固醇血症人群表现为中、重度症状/证候(后验隐类概率分别为33%和67%),证候要素以气虚、气滞、火热、痰湿、瘀血、阴虚为主,证候要素靶位以心、肝、脾、肾为主。结论隐结构模型不仅直观显示各显变量(中医症状)与隐变量(证候要素)的隐树结构,且可以定量变量的关联程度及发生概率,是较为理想的证候要素评价方法。Objective To review the characteristics of major syndrome factors of hypercholesterolemia. Methods The latent tree structure chart of 419 cases of hypercholesterolemia was established by applying multidimensional hierarchical clustering, and quantitative data, including latent probability, conditional probability, mutual information and cumulative information coverage rate, were analyzed. The characteristics of major syndrome factors of hypercholesterolemia were interpreted by using artificial recognition method. Results When taking cumulative information coverage rate reached 95% as the criterion, 419 cases had moderate and severe symptoms/syndromes （latent probability was 33% and 67% respectively）. The syndrome factors were mainly qi deficiency, qi stagnation, fire-heat, phlegm-damp, static blood and yin deficiency, and the targets of syndrome factors were mainly heart, liver, spleen and kidney. Conclusion A latent structure model can directly display the latent tree structures of observed variables （symptoms） and latent variables （syndrome factors）, as well as quantify the correlation and occurrence probability of variables, and it is an ideal method to review syndrome factors.

关 键 词：隐结构模型 证候要素 高胆固醇血症 

分 类 号：R241[医药卫生—中医诊断学]