Abstract:
In this work, we created an accurate and robust model for predicting chronic
kidney disease (CKD) in a difficult situation with three majority classes. This
helps patients avoid complications and improves their health. To achieve this
objective, we employed efficient machine learning methods (K-Nearest-Neighbors,
Decision Trees, Random Forest, Support Vector Machine, XGBoost, Logistic
Regression) and retrieved data from medical records of the nephrology consulting service in Bejaia. The classifier's performance was compared based on
accuracy rate. The Random Forest application achieved the greatest classification rate of 74% using the train/test evaluation approach.
