基于无创诊断指标构建乙型肝炎肝硬化上消化道出血的预测模型研究

目的 探讨乙型肝炎（以下简称乙肝）肝硬化患者发生上消化道出血的危险因素，并建立无创预测模型。方法 回顾性分析2019年1月—2022年12月山西医科大学第一医院收治的142例乙肝肝硬化患者的临床资料，利用Lasso回归筛选出有效预测因子，基于Logistic回归算法建立列线图预测模型，通过Bootstrap重抽样法对模型进行内部验证，采用受试者工作特征（ROC）曲线、校准曲线（CA）和决策曲线分析（DCA）评价模型，并将结果可视化。结果 142例乙肝肝硬化患者发生上消化道出血100例。经Lasso回归筛选的最佳建模指标为：性别、血红蛋白、中性粒细胞百分比、血糖、脾脏长径、门静脉内径。ROC曲线显示，列线图模型的敏感性为96.0%，特异性为83.0%,ROC曲线下面积为0.969(95%CI:0.946,0.993)，高于终末期肝病模型（MELD）评分的0.592(95%CI:0.487,0.698)和肝功能Child-Turcotte-Pugh(CTP)评分的0.623(95%CI:0.509,0.738);CA曲线提示模型的预测概率与实际概率具有较高的吻合度；DCA曲线提示使用列线图...

Construction of prediction model for upper gastrointestinal bleeding in patients with hepatitis B cirrhosis based on non-invasive indicators

Objective To investigate the risk factors for upper gastrointestinal bleeding in patients with hepatitis B-related cirrhosis (hereafter referred to as HBV-related cirrhosis) and establish a non-invasive predictive model.Methods Clinical data of 142 patients with HBV-related cirrhosis admitted to the First Hospital of Shanxi Medical University from January 2019 to December 2022 were retrospectively analyzed. Lasso regression was used to select effective predictive factors, and a logistic regression algorithm was employed to establish a nomogram predictive model. Internal validation of the model was performed using the bootstrap resampling method. The model was evaluated using receiver operating characteristic (ROC) curves, calibration curves (CA), and decision curve analysis (DCA), and the results were visualized.Results Among the 142 patients with HBV-related cirrhosis, 100 cases experienced upper gastrointestinal bleeding. The optimal modeling indicators selected by Lasso regression were gender, hemoglobin, neutrophil percentage, blood glucose, spleen longitudinal diameter, and portal vein diameter. The ROC curve showed that the sensitivity of the nomogram model was 96.0%, the specificity was 83.0%, and the area under the ROC curve (AUC) was 0.969 (95% CI: 0.946, 0.993), higher than the MELD score of 0.592 (95% CI: 0.487, 0.698) and CTP score of 0.623 (95% CI: 0.509, 0.738). The CA curve indicated good agreement between the predicted and actual probabilities of the model, and the DCA curve suggested that the use of the nomogram model could increase the net benefit for patients.Conclusion The nomogram model constructed based on gender, hemoglobin, neutrophil percentage, blood glucose, spleen longitudinal diameter, and portal vein diameter has good predictive efficacy and clinical application value for predicting upper gastrointestinal bleeding in patients with HBV-related cirrhosis.