Machine learning based potentiating impacts of 12-lead ECG for classifying paroxysmal versus non-paroxysmal atrial fibrillation

Table 3 Beta-coefficients, OR (95% CI), and p values of multivariate logistic regression for non-PAF compared to PAF. Eight variables listed above were selected by Lasso regression

	Univariate analysis**			Multivariate analysis**
Variables*	β-coefficient	OR (95% CI)	p value	β-coefficient	OR (95% CI)	p value
Left Atrium Anterior–Posterior diameter	1.20	3.31 (2.45–4.49)	< .001	0.74	2.11 (1.46–3.05)	< .001
Left Atrium volume	1.29	3.63 (2.68–4.90)	< .001	0.83	2.28 (1.59–3.29)	< .001
Fibrillatory wave amplitude (Lead II)	− 1.11	0.33 (0.24–0.44)	< .001	− 0.33	0.72 (0.48–1.08)	.110
Fibrillatory wave amplitude (Lead III)	− 1.42	0.24 (0.18–0.33)	< .001	− 0.94	0.39 (0.27–0.58)	< .001
Fibrillatory wave amplitude (aVR)	− 0.95	0.39 (0.29–0.52)	< .001	− 0.37	0.69 (0.48–0.98)	.040
RR interval (V4)	0.79	2.20 (1.64–2.94)	< .001	0.27	1.31 (0.92–1.86)	.131
Spectral Power Ratio (Lead 1)	0.55	1.73 (1.30–2.31)	< .001	0.33	1.39 (1.00–1.93)	.054
Spectral Power Ratio (V2)	0.51	1.66 (1.24–2.21)	< .001	0.40	1.50 (1.08–2.08)	.016

Area under the receiver operating characteristic Curve, AUC; confident interval, CI; electrocardiography, ECG; non-paroxysmal atrial fibrillation, non-PAF; paroxysmal atrial fibrillation, PAF

ISSN: 2466-1171