Kidney function deterioration is a progressive and often silent health issue that presents a significant challenge for global healthcare systems due to its late diagnosis and severe long-term consequences. To tackle this problem, we propose an intelligent prediction framework based on machine learning algorithms to support early detection. Our approach integrates clinical and biochemical parameters, including the estimated glomerular filtration rate (eGFR), which was calculated using the MDRD formula to enrich the dataset and improve diagnostic relevance. We employed multiple classification models: AdaBoost, Random Forest, and Support Vector Machine (SVM), and applied the Boruta feature selection algorithm to identify the most influential variables. The experimental results showed that AdaBoost achieved the best performance, reaching an accuracy of 99.1% when combined with Boruta-selected features. This demonstrates the importance of both refined feature selection and relevant biomarkers like eGFR in enhancing predictive performance for chronic kidney disease (CKD).