Towards Sustainable and Highly Accurate Prediction of Chronic Kidney Disease: A Comparative Study of Supervised Machine Learning Models

Chronic Kidney Disease (CKD) is a progressive and often asymptomatic condition that poses a significant global health burden, affecting over 850 million individuals worldwide. Early detection is critical to prevent irreversible renal damage; however, traditional diagnostic approaches are time-consuming and prone to human error. This study proposes a lightweight and computationally efficient machine learning framework for the early prediction of CKD using structured clinical data.The proposed methodology utilizes the UCI CKD dataset comprising 400 patient records with 24 attributes. A robust preprocessing pipeline is implemented, including missing value imputation, Z-score-based outlier detection, feature scaling, and correlation- driven feature selection to enhance data quality and reduce dimensionality. Three supervised learning algorithms—Stochastic Gradient Descent (SGD), Support Vector Machine (SVM), and Random Forest (RF)—are evaluated. To ensure model reliability and avoid overfitting, both train-test split (70:30) and 10-fold cross-validation techniques are employed. Experimental results demonstrate that the SGD classifier outperforms other models, achieving an accuracy of 98.33%, precision of 100%, recall of 97.3%, and F1-score of 98.63%. The absence of false-positive predictions highlights its clinical reliability. The proposed framework offers a scalable and real-time decision-support solution, particularly suitable for resource-constrained healthcare environments. Future work will focus on integrating multimodal data, including medical imaging, to further enhance diagnostic accuracy.