Prediction of Chronic Kidney Disease - A Machine Learning-Based Approach

In recent years, machine learning has played a substantial part in computer aided diagnosis (CAD) by utilizing algorithms to analyze medical data and support healthcare professionals in the diagnostic process. In addition to speeding up the diagnosis process, this technology development has enormous potential for raising the general effectiveness, precision, and accessibility of healthcare services. Chronic kidney disease (CKD) is a progressive medical illness characterized by a steady deterioration of kidney function. Machine learning applications in chronic kidney disease (CKD) include a wide range of aspects, including early diagnosis and risk prediction, as well as treatment refinement. Our goal in this research work is to explore an effective machine learning method for CKD occurrence prediction. More precisely, we started by using data pre-processing methods such as dimensionality reduction, outlier treatment, and missing value imputation. Second, we use several models—such as decision tree, logistic regression, random forest, gradient boosting, Gaussian naive bayes, and ridge classifier—to predict chronic kidney disease. Furthermore, we employed several techniques for fine-tuning algorithms to improve their performance. After fine-tuning the models, the performance of logistic regression, decision tree, random forest, gradient boosting, Gaussian naive bayes, and ridge classifier exhibited accuracies of 94%, 99%, 99%, 98%, 99%, and 93%, respectively. The results confirm that the machine learning models outperform all other models.