Table 1.

The three corpora used in this analysis. Due to technical limitations, fewer documents were labelled with diseases (n_d) than the total number of documents in each corpus (n_t).

Corpus	nt	nd
Clinical studiesa	177,609	147,235
Modelling literatureb	215,097	85,676
Positive control literaturec	687	244

${}^{n_t}$ total number of documents; ${}^{n_d}$ number of documents labelled with a disease.

^aclinicaltrials.gov.

^bMedline — text mining query for models.

^cBioModels database.