Evaluating Machine Learning Models for Chronic Kidney Disease Detection

The essential organ responsible for body homeostasis deals with Chronic Kidney Disease (CKD) as one of the principal mortality factors worldwide. The slow onset of CKD makes it difficult for people to discover until it is significantly advanced thus producing severe medical complications. A successful treatment of CKD relies on quick diagnosis and classification of patients. In Proposed work uses UCI CKD benchmark dataset is used for train and evaluate three machine learning models namely KNN, SVM and Gradient Boosting. The models undergo performance evaluation through the assessment of accuracy in addition to sensitivity and specificity indicators for CKD classification. Gradient Boosting proves to be the most trustworthy model for detecting CKD since it delivers $99.17 \%$ accuracy and $97.78 \%$ sensitivity with $\mathbf{1 0 0 \%}$ specificity. The accuracy score of SVM stands at $\mathbf{9 7. 5 0 \%}$ and KNN obtained $\mathbf{9 8. 3 3 \%}$. The research indicates Gradient Boosting stands as the most effective approach for detecting CKD patients among all evaluated modeling techniques.