儿童重症腺病毒肺炎风险预测模型的构建与验证

Construction and Validation of a Risk Prediction Model for Severe Adenovirus Pneumonia in Children

摘要: 目的：分析儿童腺病毒肺炎(AVP)的临床特征及实验室指标，构建并验证重症腺病毒肺炎的风险预测模型。方法：纳入2021年8月至2024年9月西安市儿童医院呼吸内科一病区住院206例腺病毒肺炎患儿(重症组50例，轻症组156例)，收集其一般资料、临床数据、实验室检查结果、影像学数据及合并感染情况。通过SPSS25.0和R统计软件进行数据分析，采用LASSO回归筛选变量，结合多因素Logistic回归构建预测模型，并从区分能力、校准能力和临床价值三方面评估模型性能。结果：两组在年龄、性别等基线资料上无显著差异(P > 0.05)。经LASSO回归筛选出8个危险因素，多因素Logistic回归最终确定热程、淋巴细胞百分比、白蛋白、乳酸脱氢酶、胸腔积液和合并肺炎支原体感染为独立危险因素。基于这些因素构建的列线图模型，其ROC曲线下面积(AUC)为0.938，敏感度和特异度分别为96.0%和83.9%；Hosmer-Lemeshow检验(P = 0.70)表明校准度良好；决策曲线分析显示模型在较大阈值范围内具有临床获益。其中，白蛋白是最佳预测指标(AUC = 0.873，敏感度92.0%，特异度74.4%)。结论：这种简单易用的列线图可为临床医生提供直观、量化的风险评估工具，有助于临床医生对重症腺病毒肺炎进行早期诊断。

Abstract: Objective: To analyze the clinical features and laboratory indicators of adenovirus pneumonia (AVP) in children, and to construct and validate a risk prediction model for severe adenovirus pneumonia. Methods: A total of 206 children with adenovirus pneumonia admitted to the First Ward of Respiratory Medicine, Xi’an Children’s Hospital from August 2021 to September 2024 were enrolled, including 50 cases in the severe group and 156 cases in the mild group. General information, clinical data, laboratory test results, imaging data, and co-infection status were collected. Data analysis was performed using SPSS 25.0 and R statistical software. LASSO regression was used to screen variables, and a prediction model was constructed by combining multivariate Logistic regression. The model performance was evaluated in terms of discriminative ability, calibration ability, and clinical value. Results: There were no significant differences in baseline data such as age and gender between the two groups (P > 0.05). LASSO regression screened out 8 risk factors, and multivariate Logistic regression finally identified duration of fever, lymphocyte percentage, albumin, lactate dehydrogenase, pleural effusion, and co-infection with Mycoplasma pneumoniae as independent risk factors. The nomogram model constructed based on these factors had an area under the ROC curve (AUC) of 0.938, with a sensitivity of 96.0% and a specificity of 83.9%. The Hosmer-Lemeshow test (P = 0.70) indicated good calibration. Decision curve analysis showed that the model had clinical benefits within a large threshold range. Among them, albumin was the best predictive indicator (AUC = 0.873, sensitivity 92.0%, specificity 74.4%). Conclusion: This simple and easy-to-use nomogram can provide clinicians with an intuitive and quantitative risk assessment tool, which is helpful for the early diagnosis of severe adenovirus pneumonia.