Adaboost Machine Learning Based Modelling to Predict Chronic Kidney Disease Staging

The main triggers of chronic kidney disease include diabetic mellitus along with elevated cholesterol levels and blood pressure. An individuals with Kidney disease is more likely to pass away before their prime. When trying to diagnose the many disorders connected to chronic kidney disease at the earliest stages and treat the condition, physicians underwent many obstacles. A unique machine learning technique is presented in this study for the early identification and prognosis of different stages of Kidney disease. Chronic renal failure levels are determined by estimated Glomerular Filtration Rate (eGFR). Dataset is utilized with twenty- nine attributes on renal failure obtained from the UCI Machine Learning Repository. Adaboost, SVM, Stochastic Gradient Boosting (SGD), and Neural Network were all thoroughly evaluated in this study for their potential to predict different stages of risk in chronic renal disease. Information gain and gain ratio is used in AdaBoost to rank and choose key features for ensemble model by using decision tree-based learners. Crucial assessment metrics were all included in the analysis to choose a best algorithm. The results clearly demonstrate Adaboost to be the most promising model, displaying better performance with the best classification accuracy, a balanced F1 score, and an impressive AUC value. These findings highlight Adaboost's ability for accurate risk assessment and place it as the best option for predicting risk stages in chronic renal disease.