【摘要】 目的 基于LASSO回归建立慢性乙型肝炎肝硬化患者继发肝性脑病的预测模型。方法 选择2022年1月至2024年1月在阜阳市第二人民医院接受治疗的慢性乙型肝炎肝硬化患者189例，其中继发肝性脑病的患者61例。收集患者的基线资料，比较两组基本资料和生化指标，通过LASSO-logistic回归模型筛选乙型肝炎肝硬化患者继发肝性脑病的风险因素，评价模型拟合效果，采用校准曲线验证。结果 继发肝性脑病患者的年龄、ALT、AST、TBil、ALP、血氨和MELD评分分别为(56.2±7.2)岁、(32.15±4.51)U/L、(46.72±5.48)U/L、(40.01±5.53)μmol/L、(95.67±10.28)U/L、(51.03±7.34)μmol/L、(10.13±2.54)分，均高于无肝性脑病患者的(53.0±4.5)岁、(28.83±4.10)U/L、(40.88±5.11)U/L、(28.78±5.11)μmol/L、(89.14±9.77)U/L、(30.28±5.93)μmol/L、(7.23±8.205)分，而Hb和Alb水平更低，差异均有统计学意义(P<0.05)。根据LASSO-logistic回归分析结果，高龄、高血氨、高终末期肝病模型(MELD)评分、低Alb均是患者发生肝性脑病的独立风险因素，LASSO-logistic回归模型的AIC以及BIC分别为20.221和39.672,拟合效果较好。结论 LASSO-logistic回归模型拟合效果较好，有助于提高慢性乙型肝炎肝硬化患者继发肝性脑病的预测准确度。更多还原

【Abstract】 Objective This study aims to establish a predictive model for hepatic encephalopathy secondary to chronic hepatitis B-related cirrhosis using LASSO regression analysis. Methods The study population consisted of 189 patients with chronic hepatitis B-related cirrhosis admitted from January 2022 to January 2024. Patients with hepatic encephalopathy were designated as the observation group, while the others were served as the control group. The baseline data and biochemical indices of all patients were collected and compared between the two groups. LASSO-logistic regression analysis was employed to identify the predictive factors for hepatic encephalopathy and construct a predictive mode. The model′s fit was evaluated using the akaike information criterion(AIC) and bayesian information criterion(BIC), comparing the traditional logistic and LASSO-logistic regression models, and validated by calibration curves. Results Among the 189 patients with hepatitis B-related cirrhosis, 32.28%(61/189) developed hepatic encephalopathy. The baseline data of patients in the observation group were as the following: age(56.2±7.2) years, alanine aminotransferase(ALT)(32.15±4.51) U/L, aspartate aminotransferase(AST)(46.72±5.48) U/L, total bilirubin(TBil)(40.01±5.53) μmol/L, alkaline phosphatase(ALP)(95.67±10.28) U/L, blood ammonia(51.03±7.34) μmol/L, and model for end-stage liver disease(MELD) score(10.13±2.54), all of which were higher than those of(53.0±4.5) years,(28.83±4.10) U/L,(40.88±5.11) U/L,(28.78±5.11) μmol/L,(89.14±9.77) U/L,(30.28±5.93) μmol/L, and(7.23±8.205), respectively, in patients of the control group. Additionally, the levels of hemoglobin(Hb) and albumin(Alb) were lower than those of the observation group, with statistically significant differences(P<0.05). According to LASSO-logistic regression analysis, advanced age, high blood ammonia, high MELD score, and low albumin(Alb) were identified as independent risk factors for the development of hepatic encephalopathy. The AIC and BIC values for the LASSO-logistic regression model were 20.221 and 39.672, respectively, indicating a good fit of the model. Conclusion The LASSO-logistic regression model, based on the selected variables, shows good fit and predictive accuracy for assessing hepatic encephalopathy in patients with chronic hepatitis B-related cirrhosis.更多还原