Table 4. The performance of CNN_3 and CNN_4 on the combined dataset.

The training performance on the combined dataset with different ratios collected from 5-fold cross-validation are described.

Site	Ratio	Model
		CNN_3				CNN_4
		AUPRC	Precision	Recall	MCC	AUPRC	Precision	Recall	MCC
Donor	1:1	0.987	0.944	0.969	0.841	0.988	0.940	0.973	0.846
	1:3	0.969	0.911	0.939	0.899	0.970	0.900	0.949	0.898
	1:5	0.954	0.892	0.912	0.882	0.954	0.875	0.931	0.882
	1:7	0.940	0.878	0.886	0.865	0.940	0.854	0.915	0.866
Acceptor	1:1	0.972	0.902	0.949	0.846	0.973	0.898	0.954	0.847
	1:3	0.933	0.850	0.894	0.827	0.935	0.839	0.907	0.828
	1:5	0.903	0.820	0.851	0.802	0.904	0.806	0.870	0.806
	1:7	0.877	0.804	0.811	0.780	0.878	0.788	0.835	0.783

Notes.

Bold styling emphasizes the highest values regarding the evaluation metrics used in the study.