Abstract:

Diabetes is a disease that occurs when the body does not produce sufficient amount of insulin or does not use it properly, with a blood sugar rise above the normal. Blood sugar is the main energy source of humans, and this energy comes from foods consumed. If the disease is not treated, it can be fatal. However, when early diagnosis and treatment begins, treatment is one of the most likely diseases. Since the process of diagnosing traditional diabetes is difficult, the disease can be diagnosed using systems such as artificial nerve network, image processing and deep learning using the clinical and physical data of diabetes. This study provides a deep learning-based model for the diagnosis of diabetes. In this context, the Evolutionary Neural Network (ESA) is preferred for the hybrid use of the long-term short-term memory networks (LSTM) model. Also, ESA and LSTM models have been used separately in experiments. The Pima Indians Diabetes data set, which is widely used in literature to evaluate the performance of the recommended model, has been used. The highest classification success was achieved from the ESA+LSTM model with 86.45%.

Keywords: