Chronic Kidney Disease Classification Using Machine Learning Methodology: Investigating Predictive Models

Chronic Kidney Disease (CKD) is one of the most acute health disease globally, as it generally progresses so silently until the advanced stage is reached. There is a need for timely detection and intervention for adequate management and improvement in patient outcomes. This study explores whether it is possible to accurately classify CKD using machine learning techniques. We used the two most commonly used classification algorithms-KNN and SVM-to classify a vast dataset based on several clinical parameters. The performances of the models were evaluated utilizing standard metrics such as accuracy of $\mathbf{9 5. 3 4 \%}$ for SVM and $\mathbf{9 0. 6 9 \%}$ for KNN, precision of $\mathbf{9 4. 7 3 \%}$ for SVM and $\mathbf{9 2. 4 6 \%}$ for KNN, F1-score of $\mathbf{9 5. 3 8 \%}$ for SVM and $\mathbf{9 0. 6 9 \%}$ for KNN and Recall of $\mathbf{9 6. 1 5 \%}$ for SVM and $\mathbf{9 2. 3 0 \%}$ for KNN. Our findings clearly demonstrate that SVM outperforms KNN in classification performance metrics. The superior performance of SVM can be attributed to its ability to identify optimal decision boundaries, even in complex and high-dimensional datasets. This work contributed to the growing body of literature on machine learning applications in healthcare. By applying advanced algorithms and robust evaluation techniques, we have developed a promising predictive model for classification for CKD. These findings could assist healthcare professionals in the earlier detection of those at risk of progression so that early intervention can be applied, with a corresponding better chance of influencing treatment decisions.