Table 2.

Performance metrics of the multi-label baseline classifier.

Class	Accuracy		Precision		Recall		F1-score
	B_ROC	B_PR	B_ROC	B_PR	B_ROC	B_PR	B_ROC	B_PR
Ca	0.803	0.971	0.084	0.348	0.785	0.347	0.152	0.348
Pa	0.652	0.677	0.505	0.540	0.647	0.540	0.567	0.540
Pn	0.807	0.938	0.147	0.297	0.694	0.297	0.242	0.297
Pl	0.777	0.862	0.340	0.472	0.743	0.471	0.466	0.471
Ot	0.736	0.963	0.057	0.166	0.696	0.166	0.105	0.166
MEAN	0.755	0.882	0.226	0.365	0.713	0.364	0.307	0.364

Results obtained for all instances predicted as positive for a given class, regardless of whether they are single-pathology or multi-label. The precision, recall, and F1-score values were calculated using class-specific optimal thresholds, optimized based on the ROC curve (B_ROC) or the PR curve (B_PR).