Medical Information System for Classification of Diabetes Mellitus Using Layered Neural Network

Abstract

Abstract: Diabetes Mellitus, the chronic illness condition is correlated by high blood sugar levels and troubled conditions rapidly increasing over the decades. Type 1 Diabetes occurs due to less secretion of insulin by beta cells. But beta cells produce required insulin in Type 2 diabetes where the body cannot use it. The significant factor for diabetes is insulin deficiency. Diabetes during the trimesters of pregnancy is known as Type 3 diabetes called gestational diabetes. Gestational diabetes automatically disappears after the delivery or may continue as type 2 diabetes. The research work proposes an information system for classifying all three types of diabetes Miletus using a layered neural network. The medical information system contains two major phases, Training phase, and testing phase. The architecture of the layered neural network improves the efficiency of the classification. Also, the experimental analysis and performances of diabetes classifications are obtained using the confusion matrix through Sensitivity, specificity, and accuracy. The highest value of specificity and sensitivity achieved for the proposed layered neural network is 0.96 and 0.98, respectively. The experimental results of diabetes mellitus classification exhibit 98% of high accuracy.

diabetes, gestational, classification, layered neural network, sensitivity, specificity, accuracy