Development of a deep learning algorithm for atrial fibrillation identification using single lead electrocardiogram

Published

2023 45th Annual International Conference of the IEEE Engineering in Medicine and Biology Society

Author

Sun Jung Lee, Yun Kwan Kim, Lanmi Hwang, Kyung Chul Kim, Hee Seok Song

Abstract

Since it is difficult to detect atrial fibrillation (AF) at an early stage, a method to identify the potential risk of AF using short-term electrocardiogram (ECG) monitoring is needed. In previous studies, 12-lead ECG data were generally used to
identify AF, but in this study, we used single-lead ECG data. We developed and verified two deep learning models, and selected the CNN-LSTM model that showed the best performance. As a result, our proposed method obtained an overall AUC of 0.72. This algorithm helps clinicians predict the probability of AF and design strategies for early detection and treatment for AF.