Artificial intelligence for differentiating COVID-19 from other viral pneumonias on CT: comparative analysis of different models based on quantitative and radiomic approaches

European Radiology Experimental

Table 3 Performances of Model3 and Model4 on the independent validation set

	Sensitivity	Specificity	Accuracy	AUC
*Model3*	0.788	0.783	0.786	0.834
*Model4*	0.788	0.754	0.777	0.828

Performance metrics obtained for the application of the two best models (Model3 and Model4) to the independent validation set (220 CT images). In the table are reported sensitivity, specificity, accuracy, and AUC. A threshold of 50% to the predicted value was applied to discriminated COVID and non-COVID patients. AUC area under the receiving operating characteristic curve