Computerized information system using stacked generalization for diagnosis of diabetes mellitus

Diabetes mellitus is considered to be a severe health issue which is caused due to the presence of higher amount of plasma/glucose in the blood. A number of decision support systems were introduced to help medical experts for analyzing different factors that cause diabetes. Here a computerized information system is designed using Stacked Generalization for predicting diabetes. The classifiers under consideration are Decision Stump, Decision Tree, Naive Bayes and Support Vector Machine. The selection of Stacked Generalization is done after conducting a detail performance evaluation of individual classifiers and AdaBoost algorithm. The level of accuracy was varied from a lower value of 75 % to a higher value of 82 % by using Stacking algorithm.