Fuzzy based modeling for diabetic diagnostic decision support using Artificial Neural Network

摘要

This paper investigates a variation to preliminary inquiry information obtained from patients of a diabetic and research center using fuzzy relation based model.. The proposed model is an attempt to closely replicate a physician's insight of symptom-disease associations and his approximate-reasoning for conclusion. The algorithm is evaluated on a dataset of 600 cases. The study is on people approaching diabetician with either past history of Diabetics or new case with symptoms of diabetics. Some cases are Normal patients without diabetics. The required parameters are estimated by interviewing patients. Later the parameters are modeled using a fuzzy approach and after normalization classified by Artificial neural networks as 'Close to Type 2 diabetic' or not. This result may indicate the effectiveness of proposed algorithm to optimally model the diagnostic process for small or large datasets; especially, due to its computational simplicity. Further studies on a variety of datasets in different population is required to establish such a utility.