Stacking Model for Diabetes, Heart, Liver and Kidney Disease Prediction

Abstract

Predicting chronic diseases, such as diabetes, cardiovascular diseases, and organ (liver and kidney) disease, is becoming very important in health care as it allows for early intervention and detection of diseases. It can also save lives and reduce healthcare costs to patients. Due to the combined capabilities of several algorithms (machine learning (ML) algorithms), this study examines the use of Random Forest (RF), Decision Tree (DT), and Logistic Regression (LR) alongside Gradient Boosting (GB) algorithms. The technique stacking combines different models of the algorithms and harnesses their capabilities in making a prediction and then improves classification. The preliminary results show that in respective prediction of diabetes, cardiovascular disease, and organic (liver and kidney) disease, the results of the stacking model are better in all of prediction accuracy, prediction sensitivity and prediction specificity, and greater than all other individual algorithms across all diseases.