Prediction of hearth disease using artificial neural networks

Heart disease is one of the leading daily health problems. Early diagnosis is great importance in the treatment of the disease. Artificial neural networks (ANN) is one of the artificial intelligence methods used in the disease diagnosis as in many areas. In this study, the ANN model is proposed for the diagnosis of heart disease. To develop a network that can make the most accurate diagnosis, the topology of the network should be chosen well. To determine the topology of the network in this study, the program in which the network architecture and parameters are continuously changed and the network is trained and tested, is written in MATLAB. The values ​​obtained from the program have been focused on and the architectural structure and parameters that will give the best accuracy rate have been tried. In this study, also flexible software is developed in C#.NET programming language, where the ANN model for prediction and classification problems is trained and tested in user-defined parameters. The weight values ​​of the network trained in software are stored, and it has been observed that performance is increased when the next training initial weights are performed using the weights learned. Also, the software allows the decay process for the learning rate and momentum value during training. For optimal network topology determined for heart disease, it has been observed that the accuracy rate increases when it is trained and tested by applying decay to the learning rate.