Table 2.

Efficacy of models for identifying mediastinal LNM in the training and validation groups

Model	Group	AUC	SEN	SPE	ACC	PPV	NPV
1	training	0.926	0.879	0.860	0.871	0.906	0.821
	validation	0.925	0.952	0.769	0.882	0.870	0.909
2	training	0.875	0.929	0.609	0.804	0.786	0.848
	validation	0.876	0.976	0.423	0.765	0.732	0.917
3	training	0.857	0.919	0.469	0.742	0.728	0.789
	validation	0.802	0.905	0.500	0.750	0.745	0.765
4	training	0.850	0.949	0.563	0.798	0.770	0.878
	validation	0.813	0.952	0.423	0.750	0.727	0.846
5	training	0.831	0.879	0.594	0.767	0.770	0.760
	validation	0.800	0.952	0.615	0.824	0.889	0.889
6	training	0.841	0.979	0.281	0.706	0.678	0.900
	validation	0.702	0.928	0.192	0.647	0.650	0.625

SEN sensitivity, SPE specificity, ACC accuracy, PPV positive predictive value, NPV negative predictive value