Diagnosis of Type II Diabetes Based on Feed forward Neural Network Techniques

Abstract

Diabetes is a disease caused by an increase in blood glucose levels due to insulin secretion deficiency (type I diabetes) or impaired insulin activity (type II diabetes). More than 90% of people with this condition are diagnosed with type II diabetes. Due to the sharply prevalence of type 2 diabetes in recent years, the prognosis and early diagnosis of the disease have become even more important. In this study, a model for diagnosis of type II diabetes was developed using Artificial Neural Network (ANN) method. The execution of the Frequent Pattern Growth algorithm on medical data is difficult. Association rule-based classification is an interesting area focused that can be utilized for early diagnosis. The discretization phase is necessary to transform numerical characteristics. Pima Indians Diabetes Data Set is taken as an input. The execution time, a number of rules generation and the detection of outlier percentage are analyzed. The CFP-growth algorithm utilizes for finding frequent patterns where constructing the Minimum Item Support (MIS)-tree, CFP-array and producing frequent patterns from the MIS-tree. From the set of frequent itemsets found, create all the association rules that have confidence exceeding the minimum confidence. In this study, we aim to build a model that helps Physicians in predicting Diabetes early and accurately. Data were collected from the PIMA Indian data set. Consisted of 768 samples (268 diabetic and 500 non-diabetics). Levenberg-Marquardt back-propagation algorithm was used to train the network, and the accuracy of the prediction of whether a person is diabetics or not was 88.8%.