TABLE 1.

AI‐based prediction and prognosis of DNp73 expression

CNN model	Accuracy (%)	$>$ Five years (%)	$\leq$ Five years (%)
	Biopsies without preoperative radiotherapy
ResNet50	94.00 $\pm$ 9.66	70.00 $\pm$ 48.30	100.00 $\pm$ 0.00
VGG16	94.00 $\pm$ 9.66	80.00 $\pm$ 42.16	97.50 $\pm$ 7.91
DenseNet201	96.00 $\pm$ 8.43	90.00 $\pm$ 31.62	97.50 $\pm$ 7.91
	Biopsies with preoperative radiotherapy
ResNet101	92.50 $\pm$ 12.08	100.00 $\pm$ 0.00	70.00 $\pm$ 48.30
DenseNet201	92.50 $\pm$ 16.87	96.67 $\pm$ 10.54	80.00 $\pm$ 42.16
	Tumors without preoperative radiotherapy
GoogleNet	94.29 $\pm$ 18.07	96.00 $\pm$ 12.65	90.00 $\pm$ 31.63
ResNet50	94.29 $\pm$ 12.05	96.00 $\pm$ 8.43	90.00 $\pm$ 21.08
DenseNet201	94.29 $\pm$ 13.80	96.00 $\pm$ 8.43	90.00 $\pm$ 31.62
InceptionV3	94.29 $\pm$ 12.05	100.00 $\pm$ 0.00	80.00 $\pm$ 42.16
ResNet101	95.71 $\pm$ 9.64	98.00 $\pm$ 6.32	90.00 $\pm$ 21.08
	Tumors with preoperative radiotherapy
ResNet101	90.00 $\pm$ 16.10	92.50 $\pm$ 12.08	85.00 $\pm$ 33.74
InceptionV3	93.33 $\pm$ 11.65	95.00 $\pm$ 10.54	90.00 $\pm$ 31.62
DenseNet201	93.33 $\pm$ 16.10	92.50 $\pm$ 16.87	95.00 $\pm$ 15.81
NasNetLarge	93.33 $\pm$ 16.10	95.00 $\pm$ 15.81	90.00 $\pm$ 21.08