Large-scale extraction of accurate drug-disease treatment pairs from biomedical literature for drug repurposing

BMC Bioinformatics

Table 1 Precision, recall and F1 values at different frequency cutoffs

GoldStandard	MEDLINE	Precision	Recall	F1
	Frequency
Irinotecan-Disease	>=1	0.904	0.131	0.228
	>=5	0.904	0.357	0.512
	>=10	0.904	0.509	0.651
	>=20	0.904	0.710	0.795
	>=30	0.904	0.842	0.872
Drug-Thrombocytopenia	>=1	0.800	0.186	0.302
	>=5	0.800	0.333	0.471
	>=10	0.800	0.429	0.558
	>=20	0.800	0.500	0.615
	>=30	0.800	0.667	0.727

The precisions, recalls and F1 values of extracted drug-disease pairs at different frequency cutoffs. Two evaluation datasets were used: Irinotecan-Disease and Drug-Thrombocytopenia. The precision, recall and F1 for all extracted pairs (frequency >=1) evaluated using Irinotecan-Disease dataset is 0.904, 0.131 and 0.228, respectively.

ISSN: 1471-2105