Table 2.

Overall performance in both training and validation cohorts of radiomic, clinicoradiological and combined models quantified in terms of area under the ROC curve (AUC), positive and negative predictive values (PPV, NPV), specificity and sensitivity. OR: Odds Ratio, CI: Confidence Interval.

Radomic Model		Coefficient	p-value	OR	95%CI	Overall Fit Model		AUC		95%CI		Sensivity		Specifity		PPP		NPP
	Surface To Volume Ratio	−3.82224	0.0183	0.0219	0.0009 to 0.5237	0.0097	0.0244	0.599	0.733	0.493 to 0.699	0.593 to 0.846	90.9	100.0	23.08	26.2	62.5	24.4	64.3	100.0
Clinico Radiological Model		Coefficient	p-value	OR	95%CI	Overall Fit Model		AUC		95%CI		Sensivity		Specifity		PPP		NPP
	Ca19.9	0.001128	0.049	1.0011	1.0000 to 1.0023	0.0018	0.9529	0.72	0.536	0.614 to 0.811	0.392 to 0.675	68.9	20.0	76.27	73.8	58.8	15.4	83.3	79.5
	Necrosis	1.10839	0.0633	3.0295	0.9402 to 9.7611
Combined Model		Coefficient	p-value	OR	95%CI	Overall Fit Model		AUC		95%CI		Sensivity		Specifity		PPP		NPP
	Ca19.9	0.000964	0.0946	1.011	0.9998 to 1.0021	0.0015	0.0178	0.736	0.76	0.648 to 0.838	0.618 to 0.865	58.6	60.0	86.44	78.5	68.0	40.0	81.0	89.2
	Necrosis	0.86465	0.1594	2.3742	0.7120 to 7.9167
	Surface To Volume Ratio	−2.8972	0.1171	0.0552	0.0015 to 2.0684
Training
Validation