ECG-Based Multiclass Arrhythmia Classification Using Beat-Level Fusion Network

Table 2

Comparing our work with the previous works in terms of classification performance.

Models	Macro-AUC scores
	All	Diag.	Sub-diag.	Super-diag.	Form	Rhythm
Lstm^1	0.907	0.927	0.928	0.927	0.851	0.953
Inception1d^1	0.925	0.931	0.930	0.921	0.899	0.953
Lstm_bidir^1	0.914	0.932	0.923	0.921	0.876	0.949
Resnet1d_wang^1	0.919	0.936	0.928	0.930	0.880	0.946
Fcn_wang^1	0.918	0.926	0.927	0.925	0.869	0.931
Wavelet + NN^1	0.849	0.855	0.859	0.874	0.757	0.890
Xresnet1d101^1	0.925	0.937	0.929	0.928	0.896	0.957
Ours	0.927	0.938	0.941	0.936	0.891	0.969

¹These models are stated in detail in [45]. The best performance is highlighted in bold.