Table 2.

Classification results of TIRADS and FThyNet.

		FThyNet
	TIRADS (n=2050)	Training set(n=1392)	Validation set(n=349)	Test set (n=309)
Accuracy	57.9%	79.7%	77.5%	73.0%
	[52.8–62.3%]	[77.0–82.1%]	[73.1–81.7%]	[68.2–8.1%]
Specificity	34.7%	85.9%	85.6%	79.6%
	[28.0–40.9%]	[83.0–88.8%]	[80.0–90.5%]	[73.2–88%]
Sensitivity	77.6%	73.6%	68.9%	66.7%
	[72.6–82.7%]	[69.8–77.4%]	[61.5–75.9%]	[59.4–73.5%]
AUC	56.1%	89.0%	80.1%	81.0%
	[51.8–60.3%]	[87.0–90.9%]	[75.0–84.6%]	[76.1–85.5%]

AUC, area under the receiver operating characteristic curve; TIRADS, Thyroid Imaging Reporting and Data System