Comparison of Supervised Classifiers and Strategies for Dealing with Missing Data for Chronic Kidney Disease Diagnosis

Chronic Kidney Disease (CKD) has been identified as a major global health issue since it is often asymptomatic and associated with diseases such as diabetes and hypertension. This research focuses on the need to develop better prediction models to ensure early detection and management. The goal of this study is to improve the accuracy of CKD prediction using a set of supervised machine learning methods combined with efficient missing data imputation strategies based on a dataset containing longitudinal clinical data of 10,000 patients. Thus, based on the Random Forest, Decision Tree, and Support Vector Machine algorithms, a comparative analysis is applied, which considers the usage of efficient data imputation techniques for handling missing clinical data. The experimental assessment shows that Random Forest is the best model for predicting customer churn with an average accuracy of 85% as compared to Decision Tree (79%) and Support Vector Machine (81%). Furthermore, the study also emphasizes the importance of feature selection and ensemble learning techniques for enhancing prediction reliability. These outcomes thus highlight the applicability of sophisticated machine learning algorithms in identifying and distinguishing patients in the initial stages of CKD and estimating their risk of developing further complications to allow for timely medical management. Future implications include the addition of genetic information and biomarkers of the patient to increase the level of prediction. Thus, the results of this research will help to improve the overall clinical decision-making and outcomes for CKD patients worldwide through the provision of individualized treatment regimens and more efficient utilization of healthcare resources