Development of a Predictive Analytic System for Chronic Kidney Disease using Ensemble-based Machine Learning

Chronic kidney disease (CKD) is one of the severe diseases in which kidney functions are lost gradually over months to years. Though, in the early stages of CKD little to no symptoms can be seen, but the symptoms become much more visible at the later stages of the disease and in that stage, the treatments are expensive for developing and underdeveloped countries. Again, Machine Learning (ML) models can effectively help medical professionals to predict the disease correctly in the earlier stage. Therefore, the objectives of this research are firstly, to develop an ensemble-based machine learning model to predict CKD efficiently; and secondly, to develop a predictive a nalytic system based on the proposed ensemble model. To attain these objectives, twelve ML algorithms were explored on the open-sources dataset. Then, the performance of these algorithms was analyzed to fed into the ensemble voting classifier to select the best set of algorithms. As such, seven ML algorithms were selected based on their accuracy and false negative error to be fed into the ensemble voting classifier to predict CKD efficiently. The experimental results showed that the proposed ensemble machine learning model could detect CKD with 99.17% accuracy and zero false negative error. Finally, a usable predictive analytic system was developed integrating the proposed ensemble-based ML models.