应用Logistic回归和ROC曲线探讨胃癌脾胃气虚证中医诊断模型  

Construction of TCM Diagnosis Model of Stomach Qi Deficiency Syndrome of Gastric Cancer with Application of Logistic Regression and ROC Curve

作　　者：唐继云[1,2] 李东芳[3] 侯芳芳[2] 胡亚[2] 徐佳男[2] 唐琪[2] 何寄琴 

机构地区：[1]衡阳市中医医院,湖南衡阳421099 [2]湖南中医药大学,湖南长沙410208 [3]湖南省肿瘤医院,湖南长沙410013

出　　处：《中医药导报》2016年第12期30-34,共5页Guiding Journal of Traditional Chinese Medicine and Pharmacy

基　　金：湖南省科技计划项目(NO.2014SK3044)

摘　　要：目的:应用Logistic回归和ROC曲线建立胃癌脾胃气虚证中医诊断模型。方法:收集346例胃癌患者中医四诊信息,进行证素辨证,将胃癌患者分为脾胃气虚证及非脾胃气虚证两类,探讨脾胃气虚证与常见临床症状及临床分期、红细胞计数、血红蛋白浓度、血清白蛋白、血清总蛋白浓度相关性,建立脾胃气虚证Logistic回归模型,并运用ROC曲线进行评价分析。结果:胃癌脾胃气虚证患者临床分期晚于非脾胃气虚证患者,红细胞计数、血红蛋白浓度、血清白蛋白、血清总蛋白浓度低于非脾胃气虚证患者,脾胃气虚证诊断模型:P=1/[1+exp(0.182×总蛋白浓度(g/L)-1.003×TNM分期-2.172×胃隐痛-2.846×食后腹胀-3.169×进食无味-4.822×倦怠乏力-2.281×神疲-1.630×脉细-3.669×脉缓-6.977×淡白舌)],预测函数模型整体ROC曲线下面积为0.994,灵敏度98.8%,特异度98.3%,模型检验一致率为98.6%,预测函数模型吻合率高。结论:应用Logistic回归和ROC曲线所建立的胃癌脾胃气虚证中医诊断模型,吻合率高,对于提高临床脾胃气虚证辨证准确性具有一定的参考价值。[Abstraet] Objective： To establish a TCM diagnosis model of stomach Qi deficiency syndrome of gastric cancer with applieation of logistic regression and ROC curve. Methods： The TCM four diagnostic infi）rmation of 346 gastric cancer patients were collected for syndrome element differentiation. The gastric cancer patients were divided into spleen and stomach qi deficiency and non deficiency of spleen stomach qi deficiency in 2 class. Then discuss the relativity between the common clinical symptoms and clinical staging, red blood cell count, hemoglobin concentration, serum albumin, serum total protein concentrations, and establish the spleen and stom- ach qi deficiency syndrome logistic regression model, and use the evaluation of ROC curve analysis. Results： The clinical stage of spleen stomach qi deficiency in patients with gastric cancer was later than patients with non spleen stomach qi deficiency. Red blood cell count, hemoglobin concentration, serum albumin, serum total protein eoncentration was significantly lower than that of patients with non spleen ＇stomach qi deficiency. Spleen stomach qi defieiency syndrome diagnosis model： P=l/[l＋exp （0.182~total protein concentration （g/L）-I.003~TNM staging of gastric pain-2.172~stomach pain-2.846~abdominal distension after eating-3.169~eating tasteless- 4.822~fatigue-2.281~spiritlessness-l.630~fine pulse-3.669~slow pulse-6.977~pale tongue）], prediction of area of tunction model of integrated ROC curve was 0.994, sensitivity 98.8%, specificity 98.3%, consistent rate of model test is 98.6%, the prediction function model is in high agreement. Conclusions： The model of TCM diagnosis of gastric cancer spleen stomach qi deficiency is established by the application of Logistic regression and ROC enrve.The match rate of the model is high. It has certain reference value for improving the accuracy of clinical spleen stomach qi deficiency syndrome. [Key words] gastric cancer; spleen and stomach qi deficiency; the Chinese medicine diagnosis model; l.o- gis

关 键 词：胃癌 脾胃气虚证 中医诊断模型 LOGISTIC回归 ROC曲线分析 

分 类 号：R273[医药卫生—中西医结合]