An Enhanced Method For Diabetes Prediction Using Machine Learning Approaches

Abstract

Most of the people in recent years from many parts of the world are affected with kidney failure, blindness, stroke, heart attack, etc. All such things are caused due to diabetes disease.  Because of improper creation and release of insulin in human body, several metabolism activities may go wrong, which in turn onset of diabetes. General classification of diabetes includes Type 1, Type 2 which are commonly occurred and Gestational diabetes during pregnancy. To detect and predict the disease in earlier stages becomes necessary to safe guard the people.  The machine learning approaches have been proposed to predict diabetes using PIMA dataset. With the consideration of essential features of datasets using feature selection algorithm like genetic algorithm, the performance metric like accuracy can be improved by using several machine learning algorithms. The analysis of results shows that 96.23% accuracy by using Multilayer Perceptron approach.