Comparative Analysis of Machine Learning Classifiers for Early Prediction and Diagnosis of Renal Disease

Chronic Kidney Disease (CKD) was among the most prevalent serious conditions affecting human health. Early diagnosis of CKD, a condition that often remained asymptomatic in its early stages, was crucial for preventing its progression and reducing associated mortality rates. This necessitated more intelligent techniques such as those offered by artificial intelligence (AI) in the domain of chronic disease prediction, particularly in asymptomatic early stages. Our study aimed to assist healthcare professionals in predicting and diagnosing CKD early by comparing and analyzing four machine learning models to determine the optimal model. These models included Random Forest (RF), Extra Trees (ET), Support Vector Machine (SVM), and Bagging Classifier, following the utilization of effective and precise data preprocessing and the RELIEF feature selection method. Through a comparative analysis, we concluded that the RF classifier performed the best, achieving an accuracy of 99.2% using RELIEF feature selection for prediction and early diagnosis of this disease.