Identification and validation of an explainable prediction model of favorable outcome under integrative medicine treatment exposure in DKD adult patients: a retrospective cohort study - Report - MDSpire
Advertisement
Identification and validation of an explainable prediction model of favorable outcome under integrative medicine treatment exposure in DKD adult patients: a retrospective cohort study
Clinical Report: Predictive Model for Favorable Outcomes in Adult DKD Patients
Overview
This study developed and validated a predictive model for favorable outcomes in adult patients with diabetic kidney disease (DKD) undergoing integrative medicine treatment (IMT). Utilizing a cohort of 7,400 patients, the model demonstrated an area under the curve (AUC) of 0.783 in training and 0.762 in validation.
Background
Diabetic kidney disease (DKD) affects a significant portion of adults with diabetes and is a leading cause of end-stage kidney disease. Current predictive models often lack the specificity needed to guide treatment decisions effectively.
Data Highlights
Model
AUC
Training Set
Test Set
Validation Set
XGBoost
0.783
0.715
0.762
Key Findings
XGBoost classifier was the most effective model with an AUC of 0.783 in the training set.
The model identified 10 key variables influencing outcomes: creatinine, uric acid, age, red blood cell count, urea, glucose, platelet count, calcium, white blood cell count, and sodium.
The model was validated using a temporal cohort of 3,500 patients.
The predictive model was deployed as an interactive web application for real-time predictions.
Clinical Implications
The developed predictive model can assist clinicians in identifying DKD patients who are likely to benefit from integrative medicine treatment. This approach may enhance personalized treatment strategies and improve patient outcomes in DKD management.
Conclusion
The predictive model for favorable outcomes under IMT exposure in DKD patients has been validated and is available for real-time application.
