A Novel Rule Pruning Ensemble Learning Approach to Recognize the Risk of Impaired Glucose Tolerance

Abstract: Objective: Prediction of diabetes in patients by reducing the size of rule set. Accurate prediction of diabetes risk level Methods: Early detection of diabetes can help the patients by providing information about treatment and clinical suggestions to prevent the particular individual from dangerous condition. Machine learning techniques are thus developed for accurate diagnosis of diabetes. Support vector machine (SVM) with ensemble learning approach is used for rule extraction. A novel rule pruning ensemble learning approach using frequent patterns is implemented to reduce the rule sets and improve the diagnostic performance by recognizing the risk of impaired glucose tolerance. Finding: Diabetes is a chronic condition causes high blood sugar levels. The diabetic patients are classified into type 1 and type 2diabetes. Diabetes mellitus of type 2 is considered to be the most critical worldwide public health problems that increase the level of sugar in the blood. Application/improvements: this approach accuracy, precession, recall, F-measure of the prediction of diabetes is increased.

Keywords: Diabetes mellitus, Ensemble Learning, Rule pruning, Impaired glucose tolerance.