Optimized Random Forest Model for Chronic Kidney Disease Classification with Imbalanced Data Handling

Abstract: Chronic Kidney Disease (CKD) is a progressive medical disorder that needs to be accurately diagnosed early to avoid serious complications. The aim of this study is to classify the CKD into five stages (Healthy Kidney, Mild CKD (Stage 1-2), Moderate CKD (Stage 3), Severe CKD (Stage 4), Kidney Failure (Stage 5)) using a machine learning technique. A Random Forest (RF) classifier is chosen due to its power and ability to handle high dimension clinical data. The Synthetic Minority Oversampling Technique (SMOTE) is used to solve the imbalance of classes in the data by enhancing the representation of minority classes. Moreover, the optimization of hyperparameters is done with the help of the Grid Search with cross-validation (GridSearchCV) in order to improve the performance and the cross-validation of the model. It is tested with a large pool of demographic, physiological and biochemical data, while the proposed structure is under test. The result of the experiment reveals that the optimized model has 94.12 accuracy which indicates the effectiveness of the model in the multi-class classification of CKD. The results showed that ensemble learning, data balancing, and systematic hyperparameter tuning can be effectively applied to improve the accuracy of the prediction, and the model is applicable in clinical decision support systems and early diagnosis.

Keywords: CKD, SMOTE, Random Forest, GridSearchCV, Ensemble Learning.