基于Lasso回归构建慢性乙型肝炎患者发生肝硬化风险的列线图预测模型

A nomogram prediction model for assessing the risk of cirrhosis in patients with chronic hepatitis B based on Lasso regression

目的运用Lasso回归构建慢性乙型肝炎(CHB)患者发生肝硬化的列线图模型。方法回顾性分析2023年1月1日—11月30日哈尔滨医科大学附属第一医院1 218例CHB患者的年龄、性别、实验室检测结果和肝脏超声结果, 基于R语言caret包将患者按7∶3比例为训练集(853例)和内部验证集(365例);纳入同期哈尔滨医科大学附属第一医院群力院区诊治的CHB患者作为外部验证集(185例)。采用Lasso回归和多因素Logistic回归进行变量筛选和列线图模型的构建, 分别采用受试者工作特征(ROC)曲线、校准曲线和临床决策曲线(DCA)评价预测模型的区分度、校准度和临床效用性。结果年龄、血小板、γ-谷氨酰转移酶、前白蛋白、门静脉内径和脾脏厚度作为CHB患者发生肝硬化的预测变量(P<0.05), 基于上述指标构建列线图模型。内部验证集和外部验证集ROC曲线下面积(AUC)分别为0.934(95%CI 0.910~0.959)和0.881(95%CI 0.820~0.942)。校准曲线拟合度良好(内部验证集P=0.881;外部验证集P=0.478)。DCA显示模型临床效用较高。结论年龄、血小板、γ-谷氨酰转移酶、前白蛋白、门腔静脉内径和脾脏厚度为CHB患者发生肝硬化的风险因素, 构建的列线图模型具有良好的预测价值和临床效用性。

Objective:Employ Lasso regression to develop a nomogram model for predicting the risk of liver cirrhosis in patients with chronic hepatitis B(CHB).Methods:A retrospective analysis was conducted on age,gender,laboratory test results,and liver ultrasound results of 1218 patients diagnosed with CHB at the First Affiliated Hospital of Harbin Medical University between 1 January 2023 and 30 November 2023.Based on the R caret package,patients were divided into a training set(n=853)and an internal validation set(n=365)at a ration of 7∶3,and an additional 185 patients with CHB treated at the First Affiliated Hospital of Harbin Medical University Qunli campus during the same period was included as an external validation set.Lasso regression and multiple logistic regression were employed for variable selection and nomogram model construction.The discriminative ability,calibration,and clinical utility of the prediction model were evaluated using Receiver Operating Characteristic(ROC)curves,calibration curves,Decision Curve Analysis(DCA),respectively.Results:Age,platelet,gamma-glutamine transpeptidase,prealbumin,portal vein diameter,and spleen thickness were selected as predictive variables for the occurrence of liver cirrhosis in CHB patients(P<0.05),and a nomogram model was constructed based on the aforementioned variables.The AUC values for the ROC curves in the internal validation set and external validation set were 0.934(95%CI 0.910-0.959)and 0.881(95%CI 0.820-0.942),respectively.The fitting degree of calibration curve was observed in both sets(Internal validation set P=0.881;External validation set P=0.478).DCA curves demonstrated the high clinical utility of the model.Conclusion:Age,platelet,gammaglutamine transpeptidase,prealbumin,portal vein diameter,and spleen thickness were risk factors for the occurrence of liver cirrhosis in CHB patients.The constructed nomogram model exhibits good predictive value and clinical utility.