Chronic Kidney Disease prediction using machine learning algorithms and XAI approach

This study aimed to develop a machine learning model for early diagnosis and prediction of chronic kidney disease (CKD). By employing support vector machine (SVM), random forest (RF), decision tree (DT), multi-layer perceptron (MLP), and k-nearest neighbors (KNN) algorithms, combined with permutation feature importance for explainability, we sought to improve decision-making in kidney disease management. Our findings indicate that the DT algorithm outperformed others in predicting CKD, with specific gravity, serum creatinine, hemoglobin, and diabetes mellitus identified as key predictors. This model holds potential for efficient patient screening and can be applied to more complex clinical data.

کلیدواژه ها:

chronic kidney disease ، classification algorithm ، decision tree ، explainable artificial intelligence ، early diagnosis ، permutation feature importance