Support Vector Machine Model for Enhancing the Accuracy of Chronic Kidney Disease Risk Prediction

A progressive deterioration in kidney function over time is a hallmark of kidney disease, often known as Chronic Kidney Disease (CKD). This condition is a serious illness that, if not treated, can result in kidney failure and death. To prevent further deterioration, it is essential to detect and intervene early. This research examines the identification of kidney disease through Machine Learning models applied to a dataset comprising 400 instances and 25 features. Detecting kidney disease entails recognizing indications of CKD based on medical parameters like serum creatinine levels, blood pressure and red blood cell counts. The preprocessing of data involved addressing missing values, applying binary encoding to categorical attributes, and selecting features. Four models were evaluated: the Support Vector Machine (SVM)attained the maximum efficiency of 99.17%, followed by the K-Nearest Neighbors (KNN) at 98.33%, the Random Forest (RF) at 97.50%, and the Naïve Bayes (NB) at 63.33%. The findings underscore SVM's superior efficiency in kidney disease prediction, whereas Naïve Bayes faced challenges due to its reliance on the assumption that features are independent. This work is important in the medical field because it helps detect kidney disease early, thereby lowering the risk of serious complications through prompt intervention. Diagnosis based on machine learning can help healthcare professionals enhance the accuracy of their predictions, improve patient outcomes, and finetune treatment plans.