【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。更多还原

【Abstract】 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 degree and occurrence probability of variables,and it is an ideal method to review syndrome factors.更多还原