湘西地区小儿急诊肠套叠复发预测模型的建立和效果评价

Establishment and Effect Evaluation of Predictive Model for Recurrence of Intussusception in Pediatric Emergency Medicine in Western Hunan Region

摘要: 目的：分析肠套叠空气复位后复发的危险因素，并建立预测模型。方法：2022~2023年我院明确诊断为肠套叠的病人396例，并随机将70%的病人纳入建模组(276例)，30%的病人纳入验证组(120例)。进行单因素及多因素Logistic回归分析时，我们使用了SPSS25.0软件来鉴定潜在的危险因素。使用R 4.2.1软件构建了Nomogram预测模型，通过绘制受试者操作特性(ROC)曲线并计算其曲线下面积(AUC)，评估了模型的区分能力。同时，通过绘制校准曲线，我们进一步评价了模型的校准精度。结果：总体复发率为12% (48/396)，多因素分析显示套块直径，肠壁血流信号，腹腔积液，血便是肠套叠空气复位后复发的危险因素，Nomogram预测模型的AUC = 0.82% (95% CI 74.3~89.9)，预测具有一定的预测能力。结论：在本研究中，结合超声影像学与临床数据所构建的列线图预测模型展现了良好的校准性和区分能力，为临床实践提供了有效的理论依据。

Abstract: Objective: To analyze the risk factors for recurrence of intussusception after air repositioning and to establish a predictive model. Methods: There were 396 patients with a definite diagnosis of intus-susception in our hospital from 2022 to 2023, and 70% of them were randomly included in the modeling group (276 patients) and 30% in the validation group (120 patients). For performing univariate and multivariate logistic regression analysis, we used SPSS 25.0 software to identify po-tential risk factors. A Nomogram prediction model was constructed using R 4.2.1 software, and the discriminatory ability of the model was assessed by plotting the subject operating characteristic (ROC) curves and calculating their area under the curve (AUC). Meanwhile, by plotting the calibra-tion curve, we further evaluated the calibration accuracy of the model. Results: The overall recur-rence rate was 12% (48/396), and the multifactorial analysis showed that the diameter of the tro-car mass, the blood flow signal of the intestinal wall, the abdominal fluid, and the hematocrit were the risk factors for the recurrence of the intestinal trocars after air repositioning, and the AUC of the Nomogram prediction model was = 0.82% (95% CI 74.3~89.9), which predicted a certain predictive ability. Conclusion: In this study, the column-line graph prediction model constructed by combining ultrasonography and clinical data demonstrated good calibration and differentiation ability, providing an effective theoretical basis for clinical practice.