On the Diagnosis of Diabetes Mellitus Using Artificial Neural Network Models

Diabetes mellitus is a chronic disease which occurs when the pancreas does not produce sufficient insulin, or when the body cannot effectively use the insulin it produces. It is an important and relatively common medical condition and is a risk factor for many other medical conditions like stroke, peripheral vascular disease and coronary artery disease. Physicians have to elicit a comprehensive medical history and thorough physical examination before diabetes mellitus can be suspected. In this process a lot of data has been collected on the diseases diagnosis and treatment. In this work Artificial Neural Network models were developed using both classification and predictive neural networks for the rapid diagnosis of diabetes mellitus. Both neural network models were able to learn the problem with the predictive network giving a better performance of 84% correctly classified records as opposed to 76% achieved by the classifier network on the same data set. A combined Diagnosis and Treatment neural network was also modeled using various neural network architectures. The GRNN/PNN network gave the best result out of the three architectures used. The other networks were unable to model the problem. The ultimate intention is to assist medical workers in the diagnosis process using physically measurable parameters (symptoms). Keyword: Diabetes Mellitus, Diagnosis, Artificial Neural Networks, Classification, Prediction