TABLE 5.

The performances of models on THCA and lung cancer samples.

	THCA (170) a				Lung (184)
	${\text{M}}_{\text{T}}^{\text{LOG}}$ b	MCAT	MCNN	${\text{M}}_{\text{T}}^{\text{LION}}$	${\text{M}}_{\text{L}}^{\text{LOG}}$	MCAT	MCNN	${\text{M}}_{\text{L}}^{\text{LION}}$
ACC	0.651	0.693	0.753	0.872	0.659	0.698	0.732	0.818
SEN	0.444	0.488	0.418	0.775	0.552	0.625	0.538	0.730
SPE	0.800	0.827	0.973	0.957	0.712	0.750	0.869	0.882
AUC	0.600	0.692	0.724	0.974	0.680	0.736	0.753	0.899

ACC, accuracy; SEN, sensitivity; SPE, specificity; AUC, area under the receiver operating characteristic curve; THCA, thyroid cancer.

^aEach group of samples was a mix of cancer and control samples (n = 260).

^bThe threshold of both M^LOG and M^LION was 0.5 for two samples and the thresholds of M^CAT and M^CNN were set by Youden index (M^CAT: 0.336 for THCA and 0.321 for lung cancer; M^CNN: 0.433 for THCA and 0.419 for lung cancer).