【摘要】 目的 构建HIV/AIDS患者接受cART后免疫重建不良的预测模型并验证，为临床工作者识别免疫重建不良高危人群提供参考。方法 数据来源于国家艾滋病综合防治信息系统贵州省数据，选取2016年至2018年的研究对象并按照7∶3的比例随机划分为训练集（n=11 329）和测试集（n=4 855）,2019年的研究对象作为验证集（n=7 147）。在训练集中使用LASSO回归和Logistic逐步回归筛选预测变量，然后进行多因素Logistic回归分析并构建列线图预测模型。通过绘制ROC曲线、Calibration校准曲线和决策曲线分析对模型进行评价，并根据ROC曲线找到最佳风险阈值以划分高风险人群。结果 2016年至2018年的16 184例HIV/AIDS患者中，免疫重建不良发生率为36.05%(5 834例)。多因素Logistic回归分析结果显示，性别、年龄、疾病阶段、受教育程度、感染途径和基线CD4细胞计数是HIV/AIDS患者免疫重建不良发生的相关因素（P均<0.05）。构建的预测模型最佳风险阈值为0.300，风险阈值评分为171分，模型在训练集、测试集和验证集中的AUC值分别为0.800(95%CI:0.792～0.809),0.794(95%CI:0.781～0.807),0.811(95%CI:0.801～0.821),Calibration校准曲线表明模型一致性较好，决策曲线分析表明模型有较好的净收益。结论 本研究基于性别、年龄、疾病阶段、受教育程度、感染途径和基线CD4细胞计数构建的免疫重建不良预测模型具有较好的预测性能和临床使用价值，可辅助艾滋病医护人员识别免疫重建不良的高危人群。更多还原

【Abstract】 Objective This study aimed to construct and validate a predictive model for poor immune reconstitution in patients with human immunodeficiency virus(HIV)/acquired immune deficiency syndrome(AIDS) after combination antiretroviral therapy(cART), providing a reference for identifying high-risk populations. Methods Data were obtained from the Guizhou Province dataset of the National AIDS Comprehensive Prevention and Control Information System. Study subjects from 2016 to 2018 were randomly divided into a training set(n=11 329) and a test set(n=4 855) in a 7∶3 ratio, whereas subjects from 2019 were used as a validation set(n=7 147). Least absolute shrinkage and selection operator regression and logistic stepwise regression were applied to screen predictive variables in the training set, followed by multivariate logistic regression analysis to construct a nomogram prediction model. The model was evaluated using receiver operating characteristic(ROC) curves, calibration curves, and decision curve analysis. The optimal risk threshold was determined from the ROC curve to classify high-risk populations. Results Among 16 184 patients with HIV/AIDS from 2016 to 2018, the incidence of immune reconstitution failure was 36.05%(5 834 cases). Multivariate logistic regression analysis identified sex, age, disease stage, educational level, route of infection, and baseline CD4+ T-cell count as independent risk factors for immune reconstitution failure(all P<0.05). The optimal risk threshold for the predictive model was 0.300, corresponding to a threshold score of 171 points. The area under the curve values were 0.800 [95% confidence interval(CI): 0.792-0.809] for the training set, 0.794(95%CI: 0.781-0.807) for the test set, and 0.811(95%CI: 0.801-0.821) for the validation set. The calibration curve demonstrated good consistency, and decision curve analysis indicated favorable net benefit. Conclusions The predictive model for immune reconstitution failure, based on sex, age, disease stage, educational level, route of infection, and baseline CD4+ Tcell count, demonstrated strong predictive performance and clinical utility. This model may assist healthcare professionals in identifying high-risk populations with HIV/AIDS who are at risk of poor immune reconstitution.更多还原