Table 2.

Performance comparison of different models for predicting lymph node status.

Class	Methods	AUC	ACC	SEN	SPE
Negative	Tabular	0.770(0.737–0.804)	0.723(0.693–0.752)	0.791(0.662–0.845)	0.649(0.580–0.769)
	MIL-WSI	0.709(0.672–0.746)	0.669(0.637–0.703)	0.593(0.458–0.728)	0.757(0.617–0.874)
	MMMI	0.809(0.779–0.840)	0.751(0.720–0.779)	0.768(0.616–0.855)	0.734(0.637–0.874)
ITCs	Tabular	0.619(0.501–0.738)	0.701(0.265–0.938)	0.600(0.240–0.960)	0.705(0.241–0.962)
	MIL-WSI	0.531(0.424–0.639)	0.346(0.230–0.938)	0.880(0.200–1.000)	0.329(0.205–0.964)
	MMMI	0.634(0.519–0.749)	0.746(0.392–0.880)	0.600(0.320–0.960)	0.751(0.375–0.897)
Micrometastasis	Tabular	0.636(0.582–0.690)	0.538(0.351–0.743)	0.770(0.450–0.960)	0.508(0.261–0.787)
	MIL-WSI	0.617(0.561–0.673)	0.490(0.380–0.682)	0.800(0.510–0.930)	0.440(0.302–0.706)
	MMMI	0.691(0.638–0.744)	0.623(0.431–0.773)	0.710(0.450–0.910)	0.611(0.355–0.818)
Macrometastasis	Tabular	0.748(0.710–0.785)	0.723(0.638–0.759)	0.658(0.582–0.827)	0.757(0.568–0.807)
	MIL-WSI	0.691(0.650–0.731)	0.616(0.552–0.692)	0.769(0.542–0.871)	0.544(0.415–0.747)
	MMMI	0.758(0.721–0.796)	0.734(0.647–0.773)	0.653(0.556–0.822)	0.776(0.591–0.844)