Chronic Kidney Disease Prediction Using Machine Learning Algorithms

Chronic Kidney Disease (CKD) stands as a pervasive and critical renal condition, posing a substantial global health challenge. Each kidney comprises nephrons, the essential filtration units that help to filter the blood and extract the waste products. Individuals with type 2 diabetes or high blood pressure face an elevated risk of renal failure. Predicting renal failure at an early stage remains challenging due to the absence of specific symptoms, leading to severe long-term consequences, including mortality. This paper aims to comprehensively analyze and assess the efficacy of predictive models for CKD, using various Machine Learning Classification Algorithms. The classifiers such as K-Nearest Neighbor (KNN), Decision Tree (DT), Random Forest (RF), Naïve Bayes (NB), and Gradient Boosting (GB). Out of all the classifiers Random Forest and Gradient Boosting produced the best accuracy result compared to other classifiers. This research extends the investigation by providing a nuanced analysis of the significance of machine learning techniques in predicting CKD, emphasizing their role in early detection and prognosis. This research includes a detailed comparison of accuracy scores and, a confusion matrix to identify correct and incorrect features, thereby facilitating a comprehensive examination and analysis of machine learning algorithms. The evaluation utilizes accuracy, precision, recall, and f1-score to detect the optimal model.