Construction and application of a decision tree model for children with complicated appendicitis
10.3760/cma.j.cn431274-20220713-00670
- VernacularTitle:儿童复杂性阑尾炎决策树模型的构建及临床应用价值
- Author:
Jiahu HUANG
1
;
Guoqin ZHANG
;
Quansheng YU
;
Jian LIU
;
Zhagen WANG
;
Tingjun LI
;
Lulu ZHENG
;
Zhujun GU
Author Information
1. 上海交通大学医学院附属儿童医院急诊科，上海 200062
- Keywords:
Appendicitis;
Pediatric appendicitis score;
Decision trees modell
- From:
Journal of Chinese Physician
2023;25(2):202-206,211
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To establish a decision tree model of pediatric complicated appendicitis (CA) based on Pediatric Appendicitis Score (PAS) combined with inflammatory indicators, and to evaluate its clinical application efficacy in pediatrics.Methods:The clinical data of 544 children diagnosed with appendicitis in Children′s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine from January 2018 to December 2021 was retrospectively analyzed. According to postoperative pathology, the children were divided into uncomplicated appendicitis group and CA group. The independent risk factors of CA were screened by univariate and multivariate logistic regression analysis, and these parameters were included to establish the decision tree model. The accuracy of the decision tree model was verified by receiver operating characteristic (ROC) curve.Results:Binary logistic regression analysis indicated that the PAS, C-reactive protein (CRP) and neutrophil to lymphocyte ratio (NLR) were identified as independent risk factors for complicated appendicitis in children (all P<0.05). PAS, CRP and NLR were included as covariables to construct the decision tree model and binary logistic regression model for predicting CA. The decision tree demonstrated an overall accuracy of 79.2% with a sensitivity of 86.7% and specificity of 71.9%, and achieved an area under curve (AUC) of 0.821(95% CI: 0.786-0.857). The binary logistic regression model had a sensitivity of 79.6% and specificity of 69.1%, with an overall accuracy of 75.1% and achieved an AUC of 0.808(95% CI: 0.770-0.845). Conclusions:The decision tree model based on PAS score combined with CRP, NLR is a simple, intuitive and effective tool , which can provide pediatric emergency physicians a reliable basis for diagnosis of pediatric CA.
