Table 8.

Comparison of performance of the proposed approach against other state-of-the-art algorithms for the AFIB detection task on the MIT-BIH AFIB database with the ECG segment of size 5-s.

Work	Approach	Best Performance (%)
		accuracy	ppv	sensitivity	specificity
ELP	CNN	98.17	97.78	98.57	97.76
ELP	RNN	97.93	97.63	98.24	97.61
ELP	RNN-Attention	97.96	97.87	98.08	97.84
Xia et al. [34]	SWT + CNN	98.63	-	98.79	97.87
Asgari et al. [4]	SWT + SVM	-	-	97.00	97.10
Jiang et al. [9]	RR interval irregularity + P-wave absence	-	-	98.20	97.50

ppv: positive predictive value; SWT: stationary wavelet transform