Chronic Kidney Disease Prediction Using Machine Learning: Feature Selection

Preview this chapter:

Chronic Kidney Disease Prediction Using Machine Learning: Feature Selection, Page 1 of 1

< Previous page | Next page > /docserver/preview/fulltext/9789815324099/chapter-39-1.gif

Chronic kidney disease (CKD) is a pathological condition that, if not addressed, can progress to renal failure. Machine learning models have the potential to assist in predicting chronic kidney disease by analyzing data from blood tests, urine tests, imaging tests, and biopsies. This study primarily examined a dataset of blood samples consisting of 26 patient features. These features were subsequently narrowed down to the top 10 based on their highest statistical score, which was calculated using SelectKBest. This technique enhances the accuracy and efficiency of machine learning models by reducing the dimensionality of the input data and emphasizing the most pertinent features. In this study, two approaches were examined. The K-fold crossvalidation technique achieved the greatest accuracy of 98.0%, while the average accuracy for the same technique was 96.0%. On the other hand, the Naive Bayes classifier achieved an accuracy score of approximately 93.33%. The results show promise in accurately predicting the identification of patients with chronic renal disease.