未成年发热患者流感病毒感染的风险预测模型的建立与验证

Development and validation of a model predicting the risk of influenza virus infection in underage febrile patients

目的分析未成年发热患者发生甲型流感病毒(influenza A virus,INFA)感染的独立危险因素,构建简单、有效的临床预测模型,以预测患者发生INFA感染的风险。方法选取2022年6月13日至7月5日于邵阳学院附属第一医院发热门诊就诊的未成年患者1001例,收集患者的基本资料、血常规、超敏C反应蛋白和鼻咽拭子流感病毒抗原检测结果。首先通过Lasso回归筛选预测因子,然后进行多因素Logistic回归,建立未成年发热患者发生INFA感染的风险预测模型,并用列线图展示模型。采用ROC曲线和校准曲线评价预测模型的区分度和校准度;使用决策曲线分析(DCA)评估预测模型的临床有效性。采用Boot-strap法重抽样1000次,对模型进行内部验证。结果年龄、红细胞压积(HCT)、血小板分布宽度(PDW)、血小板压积(PCT)、淋巴细胞数(Lym#)和中性粒细胞数(Neut#)是未成年发热患者发生INFA感染的风险预测因子。依据预测因子绘制列线图,构建临床预测模型。列线图模型的ROC曲线下面积(AUC ROC)为0.728(0.697~0.759),敏感性为71.89%,特异性为64.39%,约登指数为0.363,内部验证C-指数为0.720。结论构建的临床预测模型较好,可以为临床医生初步识别INFA感染提供依据。

Objective To investigate the independent risk factors of influenza A virus(INFA)infection in underage febrile patients,and construct a simple and effective clinical prediction model for predicting the risk of INFA infection.Methods 1001 underage febrile patients were recruited from the Fever Clinic,the First Affiliated Hospital of Shaoyang University from June 13,2022 to July 5,2022.Demographics,complete blood count,high sensitivity C-reactive protein levels and influenza virus antigen test results for all cases were collected.Lasso regression was applied to screen risk factors,then multivariate logistic regression analysis was used to develop the risk prediction model of INFA infection that presented with nomogram.The receiver operating characteristic(ROC)and calibration curve were used to evaluate the discrimination and calibration of the model,and the decision curve analysis(DCA)to assess the clinical validity.Bootstrap method with 1000 times resampling was used for internal validation.Results Age,hematocrit(HCT),platelet distribution width(PDW),thrombocytocrit(PCT),lymphocyte count,and neutrophil count were the risk factors for the prediction of INFA infection.The clinical prediction model was built by drawing nomogram graph based on these risk factors.The area under the ROC curve(AUC ROC)of the model was 0.728(0.697-0.759),with the sensitivity of 71.89%,the specificity of 64.39%,and the youden index was 0.363.The C-index of internal validation is 0.720.Conclusion The clinical prediction model constructed in this study is relatively accurate and concise,and can provide some help to clinicians in the preliminary identification of INFA infection.