Table 2.

Performance comparison of different models trained based on different ensemble strategies on the 10-fold cross-validation test

Ensemble strategy	Model	SN	SP	ACC	F-value	MCC
Stacking	ML model	0.836 ± 0.021	0.916 ± 0.017	0.876 ± 0.013	0.871 ± 0.015	0.755 ± 0.027
	DL model	0.623 ± 0.036	0.739 ± 0.030	0.681 ± 0.020	0.661 ± 0.024	0.365 ± 0.038
	Hybrid model	0.837 ± 0.023	0.918 ± 0.015	0.878 ± 0.014	0.872 ± 0.016	0.758 ± 0.027
Average scoring	ML model	0.847 ± 0.017	0.854 ± 0.020	0.851 ± 0.013	0.850 ± 0.014	0.701 ± 0.026
	DL model	0.616 ± 0.033	0.739 ± 0.031	0.678 ± 0.019	0.656 ± 0.022	0.358 ± 0.037
	Hybrid model	0.789 ± 0.023	0.865 ± 0.019	0.827 ± 0.015	0.820 ± 0.016	0.657 ± 0.030
Majority voting	ML model	0.845 ± 0.016	0.846 ± 0.019	0.845 ± 0.012	0.845 ± 0.013	0.691 ± 0.025
	DL model	0.595 ± 0.035	0.765 ± 0.028	0.681 ± 0.018	0.650 ± 0.023	0.367 ± 0.034
	Hybrid model	0.743 ± 0.022	0.874 ± 0.020	0.809 ± 0.016	0.795 ± 0.017	0.622 ± 0.032

Note: Values are expressed as mean ± standard deviation. The best performance value is highlighted in bold.