Table 2.

Five-fold cross-validation results of SIL models and MIL models on test datasets.

Models	AUC	Accuracy	Sensitivity	Specificity
S (S training) SIL	0.796±0.064	0.838±0.047	0.811±0.112	0.875±0.042
T (T training) SIL	0.815±0.071	0.852±0.053	0.841 ±0.092	0.863±0.070
S (Straining) MIL	0.949±0.035	0.912±0.036	0.873±0.090	0.960±0.060
T (T training) MIL	0.954±0.033	0.904±0.038	0.868±0.073	0.945±0.093
Multi-view MIL	0.961±0.026	0.925±0.060	0.873±0.120	0.986±0.032