Sanjana K.Sowmya, V.Gopalakrishnan, E. A.Soman, K. P.Atrial fibrillation is a life-threatening cardiac disease which requires a long and tedious process of detection. So, the detection of atrial fibrillation has gained great importance. One of the most reliable ways to detect cardiac disease is through analysis of ECG signal. In this paper, we show that the performance of a deep residual skip convolution neural network-based approach for automatic detection of atrial fibrillation can be improved by hyperparameter tuning. For the present work, atrial fibrillation dataset from the 2017 PhysioNet/CinC Challenge is used. The proposed method obtained an overall accuracy of 96.08% and weighted average F1 score of 0.96, a recall of 0.96 and a precision of 0.96. The main advantage of the present work is the improved accuracy achieved using a lighter model which is trained for a lesser number of epochs.
S. K., Sowmya, V., Gopalakrishnan, E. A., and Dr. Soman K. P., “Performance Improvement of Deep Residual Skip Convolution Neural Network for Atrial Fibrillation Classification”, in Evolution in Computational Intelligence, Singapore, 2021.