ECG-based multi-class arrhythmia detection using spatio-temporal attention-based convolutional recurrent neural network.

•A novel neural network model STA-CRNN is proposed by incorporating spatio-temporal attention mechanism into convolutional recurrent neural network. Spatial and temporal attention mechanisms assign weights for channels and temporal segments of a feature map, respectively. The purpose is to emphasize the informative features and suppress unimportant ones along two principle dimensions: spatial and temporal axes. Comparing with the state-of-the-art works, this study further improves the performance of arrhythmia classification.