Chronic kidney disease (CKD) is a global health concern that often progresses silently to severe complications. This study aims to enhance CKD prediction using machine learning models: support vector machines (SVM), extreme gradient boosting (XGBoost), k-nearest neighbors (k-NN), and a stacking model. The dataset, sourced from the UCI machine learning repository, includes clinical and demographic attributes from 200 patients. After preprocessing, the final dataset comprised 161 samples and 143 features. SVM achieved perfect classification performance with 100% accuracy, precision, and recall. XGBoost followed closely with an accuracy of 97.44% and a kappa statistic of 0.9451. The k-NN model delivered strong performance, achieving 92.31% accuracy. The stacking model outperformed all individual models, achieving perfect accuracy. The models demonstrated high sensitivity and specificity, indicating their effectiveness in distinguishing CKD from non-CKD cases. These findings emphasize the potential of machine learning in CKD diagnosis. Early detection can lead to improved clinical outcomes by enabling timely interventions and personalized treatment strategies. Future research should emphasize comprehensive feature engineering and larger, more diverse datasets to improve predictive accuracy and generalizability. Incorporating machine learning models in nephrology could significantly advance CKD detection and management.