International Journal of Scientific Research and Engineering Development

Chronic Kidney Disease (CKD) has become one of the most serious global health problems, affecting millions of people and increasing the burden on healthcare systems. Early identification and prevention of CKD can significantly improve patient outcomes and reduce mortality rates. This study focuses on a comparative investigation of multiple machine learning and deep learning algorithms for CKD risk prediction. Eleven models were analyzed, including traditional approaches such as Naïve Bayes, K-nearest Neighbors, Decision Tree, Random Forest, Support Vector Machine, Logistic Regression, AdaBoost, and XG Boost, along with advanced neural models such as Artificial Neural Network(ANN), Simple Recurrent Neural Network(RNN), and Long Short-Term Memory(LSTM).The experiments were performing using a CKD dataset from the UCI Repository, evaluated under three dataset conditions-unbalanced, balanced with SMOTENC, and reduced by feature selection. Each model was tested for accuracy, precision, recall, F1-score and computational efficiency to determine its effectiveness for clinical applications. The findings revealed that while most algorithms achieved comparable accuracy levels, ensemble-based methods like Random Forest, AdaBoost, and XG Boost offered a better balance between speed and performance. Deep learning models did not demonstrate notable improvements due to the datasets limited size. Overall, this research emphasizes the potential of optimized machine learning models for reliable and efficient CKD risk prediction.