DeepAR: a novel deep learning-based hybrid framework for the interpretable prediction of androgen receptor antagonists

Table 4 Performance comparison of DeepAR and conventional ML classifiers on the training and independent test datasets

Evaluation strategy	Method	ACC	Sn	Sp	MCC	AUC	F1
Cross-validation	LGBM-Circle	0.878	0.890	0.865	0.758	0.938	0.882
	MLP-All	0.886	0.907	0.862	0.774	0.934	0.891
	DeepAR	0.880	0.861	0.899	0.762	0.941	0.880
Independent test	LGBM-Circle	0.876	0.862	0.890	0.752	0.938	0.877
	MLP-All	0.888	0.874	0.902	0.776	0.949	0.889
	DeepAR	0.911	0.897	0.927	0.823	0.945	0.912

ISSN: 1758-2946