APPLICATION OF NEURAL NETS TO DETECT ARTRIAL PREMATURE BEAT ( APB )

Abstract

The atrial activity of the human heart is normally visible in the electrocardiogram ( ECG) signal as a P-wave. In patients with intermittent artery heart disease, a different P-wave morphology can sometimes be seen, indicating atrial conduction defects. The purpose of this study was to use a supervised neural network (NN)- based algorithm to discriminate between an atrial premature beats (APB) and normal ones. The performance of the method was measured using eight recordings of each type from MIT-BIH database, based on the morphology of the P-wave and the duration of the QS segment between the Q-wave of the APB beat and the previous S-wave. The method achieved a sensitivity of 92 % and a specificity of 89 % in discrimination of APB beats from normal ones. The results show that neural networks can be used in electrocardiogram (ECG) processing in cases where fast and reliable detection of artrial premature beat is desired as in the case of critical units (CCU’s).