A Deep Neural Network Ensemble Classifier with Focal Loss for Automatic Arrhythmia Classification

Table 7

Classification performance results of our method and 6 advanced methods.

Work	Acc	Class (N)			Class (S)			Class (V)
		Sen (%)	+P (%)	Spe (%)	Sen (%)	+P (%)	Spe (%)	Sen (%)	+P (%)	Spe (%)
DeChazal (2004)	85.88	99.16	86.86	94.00	38.53	75.94	95.35	81.59	77.74	98.78
Garcia (2017)	92.38	93.99	97.95	82.55	61.96	52.96	97.89	87.34	59.44	95.91
Takalo-mattila (2018)	89.91	91.89	97.00	76.83	62.49	55.86	98.11	89.23	50.85	94.02
Sellami (2019)	88.34	88.52	98.8	91.3	82.04	30.44	92.8	92.05	72.13	97.54
Jinghao Niu (2020)	95.87	98.28	97.39	78.69	77.35	73.29	98.92	85.08	91.75	99.47
Yuanlu Li (2021)	88.99	94.54	93.33	80.8	35.22	65.88	98.83	88.35	79.86	94.92
Our methods	91.89	93.15	98.18	86.00	80.23	49.40	96.85	90.99	83.09	98.72