Discrimination of β-thalassemia and iron deficiency anemia through extreme learning machine and regularized extreme learning machine based decision support system

Computer-aided systems not only assist doctors, but also minimize error rates. The purpose of these systems is to be able to think and make decisions. For this purpose, methods such as machine learning and artificial neural networks have been used. After these studies entered a period of recession due to hardware constraints, deep network research began with the development of the hardware. Deep neural networks distinguish and extract hidden data from large data by applying a back propagation algorithm for each layer. In addition, the need to extract features in deep learning methods has disappeared. The function is automatically deleted from the data set entered into the system. Another advantage is that it can be generalized. The method created for one problem can be applied to another problem. Deep learning methods require large amounts of training data to provide these benefits. When the necessary requirements for natural language processing, speech processing and image processing are met and when deep learning methods are applied, successful results and correct decisions are taken, while artificial intelligence and machine learning methods are behind. Iron deficiency anemia (IDA) is one in which iron entering the body is insufficient to produce hemoglobin. This disease is the most common type of anemia in our country and in the world. Although iron deficiency anemia (IDA) is not a genetic disease, thalassemia is an inherited feature that shortens the lifespan of red blood cells. Thalassemia is the result of abnormalities in the genes that regulate hemoglobin formation. Thalassemia screening is an attempt to find couples who may have children with thalassemia. To achieve this goal, subjects with the same type of thalassemia must be identified. It is very important to provide appropriate medical treatment for the diagnosis of the disease. Genetic diseases such as iron deficiency (IDA) anemia and-thalassemia traits require accurate and timely diagnosis. In this thesis study, in order to distinguish iron deficiency anemia (IDA) and β-thalassemia, excessive learning machines (ELM) and regular extreme learning machines (RELM) were investigated. With this study conducted to distinguish iron deficiency anemia from its-thalassemia feature, signal processing methods and machine learning methods in the literature have been concluded with software methods. As a result of this project made with the proposed method, it is seen that our method provides high performance faster and with less cost.