首页 / 院系成果 / 成果详情页

Identification and validation of an explainable prediction model of favorable outcome under integrative medicine treatment exposure in DKD adult patients: a retrospective cohort study 期刊论文

编号：

1D4D1DAE890DCFEBA26F735BE5D271AA
作者：

Jiang, Li#^[1]Zhang, Haojun(张浩军)^[2]Wang, Yanmei^[1];Yan, Meihua(严美花)*^[2]Wu, Xiai^[1];
语种：

英文
期刊：

FRONTIERS IN DIGITAL HEALTH ISSN：2673-253X 2026 年 8 卷 ; JUL 16
收录：
关键词：

diabetic kidney disease integrative medicine treatment machine learning (ML) prediction model retrospective cohort study
摘要：

Background Diabetic kidney disease (DKD) shows heterogeneous responses to integrative medicine treatment (IMT). A critical unmet need in DKD management is the inability to predict IMT response, which is essential for advancing personalized treatment strategies.Aim To develop and validate an explainable model for predicting likelihood of favorable outcome under IMT exposure in adult DKD patients.Methods A retrospective cohort comprising 7,400 patients with diabetic kidney disease (DKD) from 2010 to 2018 was analyzed. Among them, 3,900 consecutive cases diagnosed between 2010 and 2014 were randomly divided in a 7:3 ratio into a training set (n = 2,730) and an internal test set (n = 1,170), while 3,500 cases from 2014 to 2018 served as an temporal validation cohort. Feature selection was performed using the Boruta algorithm, followed by LASSO regression, Random Forest, and XGBoost-SHAP analysis. Predictive models including logistic regression, LASSO, and XGBoost were developed and evaluated based on the area under the receiver operating characteristic curve (AUC). Further assessment of model performance included sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, and the F1 score. The optimal XGBoost classifier was subsequently deployed as an interactive, single-page web application using the R Shiny framework.Results XGBoost performed best (AUC = 0.783 in training, 0.715 in test and 0.762 in validation set), with 10 key variables, namely creatinine (cr), uric acid (ua), age, red blood cell count (rbc), urea, glucose (glu), platelet count (plt), calcium (ca), white blood cell count (wbc), and sodium (Na). The web app enabled real-time prediction (https://predictionfordkd.shinyapps.io/Prediction/).Conclusion The model effectively predicts likelihood of favorable outcome under IMT exposure in DKD, aiding personalized treatment.
推荐引用方式
GB/T 7714：

Jiang Li,Zhang Haojun,Wang Yanmei, et al. Identification and validation of an explainable prediction model of favorable outcome under integrative medicine treatment exposure in DKD adult patients: a retrospective cohort study [J].FRONTIERS IN DIGITAL HEALTH,2026,8.
APA：

Jiang Li,Zhang Haojun,Wang Yanmei,Yan Meihua,&Wu Xiai.(2026).Identification and validation of an explainable prediction model of favorable outcome under integrative medicine treatment exposure in DKD adult patients: a retrospective cohort study .FRONTIERS IN DIGITAL HEALTH,8.
MLA：

Jiang Li, et al. "Identification and validation of an explainable prediction model of favorable outcome under integrative medicine treatment exposure in DKD adult patients: a retrospective cohort study" .FRONTIERS IN DIGITAL HEALTH 8(2026).
入库时间：

2026/8/20 22:29:40
更新时间：

2026/8/20 22:29:40

浏览次数：4  下载次数：0

分享到：

浏览次数：4

下载次数：0

打印次数：0