Diabetes Determination Using Retraining Neural Network

It is insulin that allows sugar to enter the cell and be stored in the cell as glycogen in the human body. Diabetes is a disease that occurs when the pancreas can not produce enough insulin or when the insulin it produces can not be used effectively. Diabetes contributes to heart health as well as kidney failure, blindness, damage to nerves and blood vessels. For this reason, the diagnosis of diabet is very important and many studies have been done for the complete diagnosis of diabetes. The work being done is usually a classificaion exercise, and it is mostly a matter of increasing the performance of the classification. In this study, a classification study for diabetes mellitus was conducted using the Pima Indian diabetes data set. For this purpose, a multilayer feed forward neural network structure trained by the Bayesian regularization algorithm and the mean square error function is used. The artificial neural network was retrained 10 times. This study was repeted 20 times. The obtained lowest all classification accuracy is 93.9% and the obtained highest all classification accuracy is 95.5%. The results of the study were compared with the results of previous which focuses on the diagnosis of diabetes studies using the same UCI machine learning dataset. The obtained results higher than the previously mentioned classification methods and evaluations show that the proposed method is very efficient and increases the classification success.