Chronic Obstructive Pulmonary Disease classification with Artificial Neural Networks

In recent years, various studies are conducted on the availability of some classifying, decisive softwares and tools that´re new and assistant upon diagnosis of the diseases. Using Artificial Neural Networks (ANN) is one of these tools. It is obvious that, these kind of systems which could be helpful for the diagnosis, and which have benefits like shortening of the diagnosis period, time gain and increased efficiency, contributes a lot to medical R&D. In this study, diagnosis of Chronic Obstructive Pulmonary Disease (COPD) with Artificial Neural Networks is objected. Dataset used in the study comprises 15 Variables, 4 COPD Disease Levels (Mild, Moderate, Severe; Very Severe), and the data of 507 patients. Within the study, (Matlab Code with 2 hidden layers); 5 layered crossvalidation method is used. Mean Squared Error (MSE) and Mean Absolute Error (MAE) of the test samples are calculated and mean of 5 layer errors are given as the results. In conclusion MSE and MAE values are found to be 0,00996, and 0,02478 respectively. Accuracy rates are found to be 99 %. Ultimately founded results show that very high accuracy rates are achieved with 15 variables. Consequently, Artificial Neural Networks seem to have a successful usage in COPD classification. This allows a system that would help the physician to determine COPD level more rapidly with high performance.