Table 1.

Performance metrics of ResNet50 and ResNet50v2 with the SAM and SGD optimizers in contact analysis.

CNN	Optimizer	Accuracy	Precision	Recall	F1 score	AUC
		SD	SD	SD	SD	SD
		95%CI	95% CI	95% CI	95% CI	95% CI
ResNet50	SAM	0.855	0.810	0.785	0.794	0.883
		0.005	0.009	0.009	0.008	0.007
		0.853–0.857	0.807–0.813	0.782–0.789	0.791–0.797	0.880–0.885
ResNet50	SGD	0.850	0.804	0.781	0.789	0.875
		0.009	0.010	0.010	0.009	0.008
		0.847–0.853	0.800–0.807	0.785–0.778	0.786–0.793	0.872–0.878
ResNet50v2	SAM	0.860	0.816	0.791	0.800	0.890
		0.005	0.008	0.009	0.008	0.007
		0.858–0.861	0.813–0.819	0.788–0.794	0.798–0.803	0.888–0.893
ResNet50v2	SGD	0.853	0.809	0.782	0.792	0.884
		0.005	0.008	0.009	0.007	0.006
		0.851–0.855	0.806–0.812	0.779–0.785	0.790–0.795	0.882–0.886

SD, standard deviation; 95% CI, 95% confidence interval; AUC, area under the receiver operating characteristics curve.