基于潜在类别结合隐结构分析的血脂异常患者中医常见证候特征研究＊

目的 探索基于潜在类别结合隐结构分析的血脂异常中医常见证候,为临床辨证提供依据。方法 多中心采集2013年1月至2016年9月1019例诊断为血脂异常患者的中医四诊信息,建立数据库。采用Mplus 8.3和Lantern 5.0软件,运用潜在类别和隐结构模型法分析法,对比分析血脂异常患者的中医常见证候、症状特征。结果 本研究选择发生率>10%的症状、体征,纳入显在变量指标34个,得到3个较为理想的数据拟合模型（BIC=37734.732）。通过潜在类别分析分为3类共6种主要病性证候要素：痰浊、气虚、血瘀、湿热、阴虚、气滞,病位证候要素以脾、肺、心、肝为主。另纳入患者的89个症状、体征建立隐结构模型,得到17个隐性变量;根据潜在类别、专家组意见,对隐性变量进行综合聚类,最终得到17个因变量、综合聚类为5个证候,包括阴阳两虚证、气滞血瘀证、痰浊阻滞证、气虚证、湿热内盛证,并初步确定了各证候的诊断要点。结论 基于潜在类别和隐结构模型分析发现血脂异常的中医常见证候为阴阳两虚证、气滞血瘀证、痰浊阻滞证、气虚证、湿热内盛证。通过建立数据模型,从显性变量中推导出隐性知识,为中医证候研究提供了新的思路,为中医治疗血脂异常的新药研发的提供了一定的科学基础。

A Study on the Characteristics of Common Chinese Medicine Symptoms in Patients with Dyslipidemia Based on Latent Class Analysis Combined with Latent Structure Model

Objective To explore common Chinese medical evidence of dyslipidemia based on latent class analysis combined with latent structure model, and to provide a basis for clinical identification.Methods The information of 1019 patients diagnosed with dyslipidemia from January 2013 to September 2016 in four Chinese medicine consultations was collected in a multicenter to establish a database. Mplus 8.3 and Lantern 5.0 software were used to compare and analyze the common TCM symptoms and symptom characteristics of patients with dyslipidemia using the latent class analysis combined with latent structure model.Results In this study, symptoms and signs with an incidence of >10% were selected, and 34 indicators of explicit variables were included to obtain 3 more ideal data fitting models (BIC=37734.732). The potential categories were analyzed into 3 categories with 6 main pathological elements: phlegm retention, qi deficiency, blood stasis, damp-heat, yin deficiency, and qi stagnation, and the pathological elements were mainly spleen, lung, heart, and liver. In addition, 89 symptoms and signs of patients were included to establish the implicit structure model, and 17 implicit variables were obtained; based on the potential categories and expert group opinions, the implicit variables were synthetically clustered, and finally 17 dependent variables and synthetically clustered into 5 symptoms, including yin and yang deficiency, qi and blood stagnation, phlegmatic blockage, qi deficiency, and damp-heat syndrome, and the diagnostic points of each symptom were initially determined.Conclusion Based on the analysis of potential categories and hidden structure model, the common TCM symptoms of dyslipidemia were found to be yin and yang deficiency, qi and blood stagnation, phlegmatic blockage, qi deficiency, and damp-heat syndrome. By establishing a data model, implicit knowledge was deduced from the explicit variables, which provided a new idea for the study of TCM symptoms and a reference for clinical evidence identification.