Prediction of diabetes mellitus based on boosting ensemble modeling

Abstract

Healthcare systems provide personalized services in wide spread domains to help patients in fitting themselves into their normal activities of life. This study is focused on the prediction of diabetes types of patients based on their personal and clinical information using a boosting ensemble technique that internally uses random committee classifier. To evaluate the technique, a real set of data containing 100 records is used. The prediction accuracy obtained is 81.0% based on experiments performed in Weka with 10-fold cross validation.