Table 3.

Performance of prediction models with and without biological tumor markers

Model	AIC	R2	Brier	AUC	DS	Int	Slope
M1	94.7	0.140	0.158	0.657	0.083	0.086	0.817
M2	97.4	0.103	0.163	0.654	0.058	0.047	0.826
M3	92.3	0.173	0.151	0.685	0.105	0.035	0.895
M4 = M1 + HER2	90.9	0.133	0.153	0.700	0.075	0.041	0.894
M5 = M2 + HER2	92.3	0.115	0.155	0.694	0.063	0.033	0.880
M6 = M3 + HER2	89.1	0.162	0.147	0.700	0.094	0.030	0.894
M7 = M1 + CD44	45.1	0.127	0.163	0.739	0.046	0.072	0.758
M8 = M2 + CD44	47.3	0.087	0.174	0.748	0.017	0.064	0.701
M9 = M3 + CD44	46.5	0.114	0.175	0.737	0.016	0.080	0.643
M10 = M1 + HER2 + CD44	40.5	0.221	0.146	0.759	0.106	0.069	0.763
M11 = M2 + HER2 + CD44	41.2	0.270	0.135	0.857	0.134	0.036	0.834
M12 = M3 + HER2 + CD44	42.0	0.225	0.150	0.816	0.073	0.060	0.724

Abbreviations: M1, clinical reference model; M2, radiomic reference model; M3, clinico-radiomic reference model; HER2, human epidermal growth factor receptor 2; CD44, cluster of differentiation 44; AIC, Akaike Information Criterion; R², Nagelkerke R²; Brier, Brier score; AUC, area under the receiver operating characteristic; DS, discrimination slope; Int, intercept