. Author manuscript; available in PMC: 2025 Oct 15.

Published in final edited form as: IEEE Access. 2025 Jul 28;13:133351–133369. doi: 10.1109/access.2025.3593420

TABLE 1.

T-test results comparing the ood auc in the (cam16,panda) experiment using uni features.

(a) Statistical comparison for vaeabmil
Model	OoD/auroc	t_stat	p_value	Significant
vaeabmil	0.9999 ± 0.0001	-	-	-
transmit	0.9690 ± 0.0059	8.9761	0.0122	True
clam	0.8300 ± 0.1985	1.4827	0.2764	False
dsmil	0.9438 ± 0.0224	4.3511	0.0490	True
dftdmil	0.9497 ± 0.0351	2.4722	0.1320	False
daeabmil	0.9923 ± 0.0001	352.6000	0.0000	True
abmil	0.9544 ± 0.0135	5.8111	0.0284	True
(b) Statistical comparison for daeabmil
Model	OoD/auroc	t_stat	p_value	Significant

daeabmil	0.9923 ± 0.0001	-	-	-
transmil	0.9690 ± 0.0059	6.7671	0.0211	True
clam	0.8300 ± 0.1985	1.4161	0.2924	False
dsmil	0.9438 ± 0.0224	3.7590	0.0640	False
dftdmil	0.9497 ± 0.0351	2.0995	0.1706	False
vaeabmil	0.9999 ± 0.0001	−352.6000	0.0000	True
abmil	0.9544 ± 0.0135	4.8494	0.0400	True