Early Stage Forcast of CKD Using ML Classifiers

Concurring to projections, the frequency of chronic kidney disease (CKD) is anticipated to extend over time. To facilitate rapid detection and cure, precise CKD prediction models are necessary. The classifier choice significantly influences the accuracy of these models. In this study, we highlight and assess various ML classification algorithms for CKD executed on dataset available on UCI library. Our study involved implementing six classification algorithms, including Random Forest, KNN, Decision Tree, Gradient Boosting, Xg Boost, and Ada Boost. In addition, we also employed random sampling and mean/mode sampling for the essential attribute selection method to refine the dataset. According to the findings, it was determined that Random Forest yielded an optimal accuracy of 98.33%. Together with precision, F-1 score, accuracy, and recall possess endured calculated and analyzed outcomes about different methods exhibited illustrated in the chart. After reviewing the analysis of the results, it is evident that the implementation of machine learning classification algorithms can prove advantageous for healthcare. Moreover, it can assist in the initial identification of CKD and related stages, leading to a reduction in the development of renal disease.