原发性自发性气胸复发预测模型的构建与验证

Development and validation of a prediction recurrence model for primary spontaneous pneumothorax

目的分析原发性自发性气胸(primary spontaneous pneumothorax,PSP)复发的危险因素,并建立预测模型。方法2010年1月~2021年1月我院明确诊断为PSP的病人803例,并随机将70%的病人纳入建模组(562例),30%的病人纳入验证组(241例)。使用R 4.2.1软件进行单因素及多因素Cox回归分析,并建立Nomogram预测模型。绘制受试者工作特征曲线,计算曲线下面积(AUC)以评估模型区分度,并绘制校准曲线以评估模型校准度。结果总体复发率为22.67%(182/803),多因素分析显示,年龄、吸烟指数、病变严重程度评分和治疗手段是PSP复发的独立危险因素,Nomogram预测模型的AUC=71.7%(95%CI 64.1~79.2),预测效能较高。结论构建的PSP复发预测模型可协助临床医生个体化评估病人复发风险。

Objective To analyze the risk factors for recurrence of primary spontaneous pneumothorax and to establish a prediction model.Methods The clinical data of 803 patients clearly diagnosed with primary spontaneous pneumothorax in the First Affiliated Hospital of Xinjiang Medical University from January 2010 to January 2021 were retrospectively analyzed,and 70%of the patients were randomly included in the modeling group(562 patients)and 30%in the validation group(241 patients).Risk factors for recurrence were analyzed by univariate and multivariate Cox regression using R 4.2.1 software,and a Nomogram prediction model was developed.Receiver operating characteristic curves were plotted,and the area under the curve(AUC)was calculated to assess model discrimination,and calibration curves were plotted to assess model calibration.Results The overall recurrence rate was 22.67%(182/803).Multivariate Cox regression analysis showed that age,smoking index,dystrophic severity score and treatment regimens were independent risk factors for recurrence of primary spontaneous pneumothorax,and the AUC of the Nomogram prediction model was 71.7%(95%CI 64.1⁃79.2),with high predictive efficiency.Conclusion This recurrence prediction model of primary spontaneous pneumothorax can assist clinicians to accurately assess the risk of recurrence in individual patients.