基于LASSO回归的非酒精性脂肪性肝病进展性肝纤维化预测模型的构建及分析  

Construction and analysis of a predictive model for progressive liver fibrosis in nonalcoholic fatty liver disease based on LASSO regression

作　　者：冯巩 张燕 严琴琴[1] 刘婧 刘曼玲[1] 范立萍 林思岑 贺娜 弥曼[1] FENG Gong;ZHANG Yan;YAN Qinqin;LIU Jing;LIU Manling;FAN Liping;LIN Sicen;HE Na;MI Man(Xi’an Medical University,Xi’an 710021,China;Department of Respiratory Medicine,Tangdu Hospital Affiliated to Air Force Medical University,Xi’an 710000,China;Department of Gastroenterology,The First Affiliated Hospital of Xi’an Medical University,Xi’an 710077,China)

机构地区：[1]西安医学院,西安710021 [2]空军军医大学第二附属唐都医院呼吸内科,西安710000 [3]西安医学院第一附属医院消化内科,西安710077

出　　处：《临床肝胆病杂志》2020年第10期2214-2218,共5页Journal of Clinical Hepatology

基　　金：佑安肝病感染病专科医疗联盟科研专项基金(LM202003);陕西省普通高等学校优势学科建设项目(陕教位〔2014〕3号文件)。

摘　　要：目的利用与非酒精性脂肪性肝病(NAFLD)相关的常用临床及实验室指标,构建无创性LASSO回归模型,并评估该模型对NAFLD进展性纤维化的预测及诊断价值。方法选择2018年1月-2019年8月在西安医学院第一附属医院连续入院的NAFLD患者258例,根据FibroScan测得肝脏组织硬度值(LSM),并依据LSM值将患者分为非进展性纤维化组184例,进展性纤维化组74例。收集研究对象的一般资料、生化指标等数据。符合正态分布的计量资料2组间比较采用t检验,不符合正态分布的计量资料2组间比较采用Mann-Whitney U检验。计数资料两组间比较采用χ2检验。使用LASSO回归算法筛选具有非零系数的进展性纤维化相关的特征指标,构建LASSO回归模型,计算模型的受试者工作特征曲线下面积(AUC)、敏感度及特异度,并将LASSO回归模型与已知的经典模型进行比较。结果通过LASSO回归,选择出的重要变量为Ⅳ型胶原、BMI、AST,据此构建LASSO回归模型。结果显示,LASSO回归模型的AUC为0.843(95%Cl:0.790~0.897),敏感度为0.851,特异度为0.810,显著优于APRI(AUC:0.791,95%Cl:0.731~0.850)、FIB-4(AUC:0.426,95%Cl:0.345~0.507)和NFS(AUC:0.540,95%Cl:0.463~0.617)。结论与目前已知的NAFLD进展性肝纤维化无创性评分系统相比,此回归模型的AUC、特异度、敏感度均较好,实用性及可操作性强,可作为NAFLD新的无创性肝纤维化诊断模型。Objective To construct a noninvasive LASSO regression model based on commonly used clinical and laboratory markers associated with NAFLD,and to investigate the value of this model in the prediction and diagnosis of progressive liver fibrosis in NAFLD.Methods A total of 258 NAFLD patients who were consecutively admitted to The First Affiliated Hospital of Xi’an Medical University from January 2018 to August 2019 were enrolled,and according to liver stiffness measurement measured by FibroScan,the patients were divided into non-progressive fibrosis group with 184 patients and progressive fibrosis group with 74 patients.General information and biochemical parameters were collected.The independent samples t-test was used for comparison of normally distributed continuous data between the two groups,and the Mann-Whitney U test was used for comparison of non-normally distributed continuous data between the two groups.The chi-square test was used for comparison of categorical data between two groups.The LASSO regression algorithm was used to screen out the characteristic indicators with non-zero coefficients which were associated with progressive fibrosis,and a LASSO regression model was constructed.The area under the receiver operator characteristic curve(AUC),sensitivity,and specificity of this model were calculated,and the LASSO regression model was compared with known classic models.Results The LASSO regression analysis screened out the important variables of typeⅣcollagen,body mass index,and aspartate aminotransferase,and a LASSO regression model was constructed based on these three indicators.The results showed that the LASSO regression model had an AUC of 0.843(95%confidence interval[CI]:0.790-0.897),a sensitivity of 0.851,and a specificity of 0.810,with a significantly better AUC than APRI(AUC=0.791,95%CI:0.731-0.850),FIB-4(AUC=0.426,95%CI:0.345-0.507),and NFS(AUC=0.540,95%CI:0.463-0.617).Conclusion Compared with the existing noninvasive scoring system for progressive liver fibrosis in NAFLD,this regression mo

关 键 词：非酒精性脂肪性肝病 肝硬化 模型 统计学 

分 类 号：R575.5[医药卫生—消化系统]