Table 4. Results from different ML algorithms.

The best ML algorithms were LSTM and SVM, whose performances are highlighted in bold.

Model	Subset	AUC	Accuracy	Recall	Precision
LSTM	Train	0.987±0.022	0.968±0.051	0.951±0.082	0.954±0.074
	Test	0.983	0.978	0.978	0.978
SVM	Train	0.951±0.020	0.936±0.026	0.936±0.026	0.940±0.025
	Test	0.946	0.928	0.928	0.928
LR	Train	0.948±0.022	0.931±0.029	0.931±0.029	0.934±0.028
	Test	0.946	0.928	0.928	0.927
CNN	Train	0.956±0.042	0.904±0.048	0.852±0.072	0.860±0.073
	Test	0.938	0.920	0.913	0.913
MLP	Train	0.932±0.025	0.910±0.033	0.910±0.033	0.913±0.035
	Test	0.929	0.905	0.905	0.905
NB	Train	0.834±0.028	0.779±0.037	0.779±0.037	0.800±0.042
	Test	0.854	0.805	0.805	0.812