Table 2.

Performance of predictive models.

Models	Datasets	AUC			Accuracy			Precision			Sensitivity			Specificity
		Clinic	Radiomic	Clinic + Radiomic	Clinic	Radiomic	Clinic + Radiomic	Clinic	Radiomic	Clinic + Radiomic	Clinic	Radiomic	Clinic + Radiomic	Clinic	Radiomic	Clinic + Radiomic
RF	Training set	0.981	0.982	0.983	0.980	0.973	0.988	0.873	0.878	0.882	0.929	0.968	0.952	0.865	0.865	0.873
	Testing set	0.721	0.746	0.879	0.787	0.787	0.863	0.428	0.556	0.778	0.500	0.417	0.583	0.851	0.926	0.963
XGB	Training set	0.948	0.957	0.960	0.936	0.952	0.968	0.828	0.824	0.801	0.762	0.929	0.960	0.841	0.802	0.762
	Testing set	0.765	0.711	0.761	0.727	0.757	0.727	0.429	0.444	0.308	0.500	0.333	0.667	0.852	0.907	0.667
LR	Training set	0.778	0.734	0.796	0.746	0.706	0.698	0.717	0.682	0.977	0.865	0.698	0.992	0.659	0.675	0.976
	Testing set	0.754	0.560	0.730	0.697	0.667	0.667	0.313	0.250	0.323	0.833	0.417	0.833	0.593	0.722	0.611
SVM	Training set	0.989	0.500	0.979	0.710	0.500	0.968	0.944	0.632	0.984	0.944	0.952	0.992	0.944	0.444	0.984
	Testing set	0.732	0.500	0.718	0.515	0.182	0.788	0.444	0.182	0.500	0.333	1.000	0.083	0.907	0.000	0.981
KNN	Training set	0.905	0.797	0.868	0.968	0.746	0.940	0.935	0.632	0.867	0.794	0.952	0.825	0.944	0.444	0.873
	Testing set	0.623	0.583	0.623	0.712	0.682	0.742	0.429	0.182	0.500	0.250	1.000	0.083	0.926	0.000	0.981

AUC, area under curve; RF, random forest; XGB, eXtreme gradient boosting; LR, logistic regression; SVM, support vector machine; KNN, k-nearest neighbors.