Research Information System
in: New Frontiers In Engineering, Akdemir Bayram,Özkaya Umut, Editor, Duvar Publishing, İzmir, pp.123-150, 2023
In this study, a novel mobile software application that can run on Android based mobile devices has been designed and developed to classify murmur heartsounds using the support vector machines method. The occurrence of heart
sounds with murmur at specific age groups and certain conditions may be symptoms of serious health problems. In order to detect this situation, an Android-based mobile application has been designed and implemented with the
JAVA programming language by considering the characteristic specialties of mobile devices. First, the heart sound signals have been normalized to develop the application. After that, low and high pass filters of Butterworth, Chebyshev-
II, and Elliptic were utilized to remove unwanted data from extra systole heart sounds. It is observed that Chebyshev-II and Elliptic filters are more successful in denoising. When all this is considered, Elliptic was selected for use in the
mobile application because it has less runtime. After that, the features vector was obtained with wavelet transformation operation. Finally, Gentle Boost, Gradient Boosting, Neural Network, and K-Nearest Neighbors classification methods have been employed on mean, variance, covariance, standard deviation, maximum value, minimum value, median value, most frequent values, detail coefficients length of features. It was determined that the classification success of the developed application was 87%, according to error matrice. The application waspresented to the opinions of cardiologists for evaluation.