Table 2.

Quantitative index comparisons of SM, RM, NTLR, and TLR on three cohorts.

Method	AUC	ACC	SENS	SPEC		PPV	NPV			MCC	F1 score
Testing set of the main cohort
SM	0.83	0.77	0.72	0.82		0.76	0.78			0.54	0.74
RM	0.64	0.62	0.71	0.57		0.50	0.76			0.27	0.58
NTLR	0.82	0.79	0.75	0.81		0.73	0.83			0.56	0.74
TLR	0.93	0.84	0.94	0.77		0.73	0.95			0.69	0.82
Independent testing set 1
SM	0.67	0.61	0.43	0.87		0.83	0.52			0.32	0.57
RM	0.55	0.51	0.36	0.72		0.65	0.44			0.08	0.47
NTLR	0.81	0.75	0.71	0.81		0.84	0.66			0.51	0.77
TLR	0.93	0.86	0.83	0.89		0.92	0.78			0.71	0.87
Independent testing set 2
SM	0.67	0.67	0.68	0.67		0.66	0.69			0.35	0.67
RM	0.57	0.60	0.47	0.69		0.53	0.64			0.16	0.50
NTLR	0.79	0.73	0.67	0.78		0.70	0.76			0.46	0.68
TLR	0.93	0.84	0.95	0.75		0.74	0.96			0.70	0.83
Significance level of Delong test for methods compared with TLR
Testing dataset		SM			RM				NTLR
Main cohort		0.0025			<0.0001			0.0012
Independent set 1		<0.0001			<0.0001			<0.0001
Independent set 2		<0.0001			<0.0001			<0.0001

The bold values represent the best value of an index in the comparative experiments.