Table 2.

Metrics for the premature ventricular contraction PVC and dual-rhythm models with differences on the algorithm development data set and external validation data set

	Accuracy (95% CI)	Precision (95% CI)	Recall (95% CI)	specificity (95% CI)	AUC_ROC (95% CI)	F1-scorea (95% CI)
Algorithm development data set
PVC model	0.78 (0.74–0.80)	0.59 (0.52–0.67)	0.79 (0.72–0.79)	0.76 (0.71–0.81)	0.88 (0.85–0.92)	0.68 (0.62–0.74)
Dual-rhythm model	0.87 (0.84–0.90)	0.92 (0.91–1.00)	0.63 (0.55–0.72)	0.98 (0.98–1.00)	0.92 (0.89–0.94)	0.75 (0.69–0.81)
Differences	0.10 (0.07–0.15)	0.34 (0.28–0.42)	−0.17 (−0.22 to −0.12)	0.23 (0.18–0.28)	0.04 (0.01–0.06)	0.07 (0.01–0.14)
P-value	< 0.01	< 0.01	< 0.01	< 0.01	0.02	0.03
External validation data set
PVC model	0.74 (0.70–0.79)	0.55 (0.47–0.63)	0.87 (0.81–0.95)	0.68 (0.62–0.75)	0.86 (0.82–0.90)	0.68 (0.60–0.74)
Dual-rhythm model	0.84 (0.81–0.89)	0.73 (0.65–0.83)	0.80 (0.74–0.90)	0.87 (0.82–0.92)	0.91 (0.88–0.95)	0.77 (0.70–0.83)
Differences	0.11 (0.06–0.16)	0.18 (0.11–0.27)	−0.05 (−0.16 to −0.04)	0.19 (0.12–0.26)	0.03 (0.01–0.09)	0.09 (0.02–0.17)
P-value	< 0.01	< 0.01	0.03	<0.01	0.03	0.03

AUC-ROC, area under the curve for the receiver operating characteristic; CI, confidence interval; PVC, premature ventricular contraction.

^aF1 score was calculated as follows: F1-score = 2 × Precision × Recall/(Precision + Recall).