An end-to-end atrial fibrillation detection by a novel residual-based temporal attention convolutional neural network with exponential nonlinearity loss

作者：

Highlights：

• RTA-CNN is proposed based on AF rhythm characteristics to adaptively enhance temporal features.

• Exponential nonlinearity loss is proposed to alleviate unbalance problems in AF detection tasks.

• On a open source dataset, the algorithm achieves competitive results compared with other SOTA methods.

• The attention mechanism of the method is appropriate for the occasional strike of paroxysmal AF.

摘要

•RTA-CNN is proposed based on AF rhythm characteristics to adaptively enhance temporal features.•Exponential nonlinearity loss is proposed to alleviate unbalance problems in AF detection tasks.•On a open source dataset, the algorithm achieves competitive results compared with other SOTA methods.•The attention mechanism of the method is appropriate for the occasional strike of paroxysmal AF.

论文关键词：Atrial fibrillation,Electrocardiogram,Convolutional neural network,Attention mechanism

论文评审过程：Received 3 March 2020, Revised 10 September 2020, Accepted 31 October 2020, Available online 11 November 2020, Version of Record 24 December 2020.

论文官网地址：https://doi.org/10.1016/j.knosys.2020.106589