Table 4.

Predictive performance of EGFR and PD-L1 status performance using three methods in the joint cohort.

Methods	Categories	AUC (95% CI)	Accuracy (%)	Sensitivity (%)	Specificity (&)
EGFR&PD-L1 Status
DL	EGFR	0.906(0.885-0.930)	0.767(0.735-0.798)	0.920(0.884-0.954)	0.787(0.748-0.827)
	PD-L1	0.879(0.854-0.906)	0.793(0.762-0.819)	0.781(0.742-0.819)	0.939(0.904-0.969)
Radiomics	EGFR	0.860(0.816-0.902)	0.659(0.617-0.705)	0.896(0.856-0.944)	0.731(0.664-0.801)
	PD-L1	0.856(0.815-0.897)	0.719(0.677-0.766)	0.713(0.661-0.763)	0.918(0.864-0.972)
Joint	EGFR	0.928(0.909-0.946)	0.831(0.807-0.856)	0.917(0.883-0.941)	0.807(0.771-0.846)
	PD-L1	0.905(0.886-0.930)	0.848(0.825-0.874)	0.847(0.818-0.879)	0.915(0.876-0.951)

EGFR, epidermal growth factor receptor; PD-L1, programmed death ligand-1; DL model, deep learning model.