Pediatric Irritable Bowel Syndrome Prediction Using 2 - Tier Ensemble Classifier

Abstract

Irritable Bowel Syndrome (IBS) is a chronic, painful digestive disease that adversely affects childs wellbeing and health. It is generally in the form of frequent abdominal twinge in child that can make life of child pathetic. In such a situation, a disease prediction model can be of huge assistance in recognizing high-risk individuals. In this work, a novel bi-level ensemble model is developed for prediction of IBS. The work employs the Wrapper method for determining the optimal set of features. In this approach of ensemble, five different models of machine learning are combined in a group of three with different permutation to further improve the results. The majority voting technique is employed to get the outcome of meta-classifiers. The novel model achieves accuracy of 92.754 % with 67:33 ratio of training test balanced class data splitter. The results revealed that the correctness of proposed model is increased in contrast to the single model accuracy. This is the first initiative to predict IBS among children. It is possible that early forecasting of chronic diseases can impact a huge number of individuals and lessen the prevalence and expenditure of these diseases.