Skip to main content
. 2008 Jan-Feb;15(1):87–98. doi: 10.1197/jamia.M2401

Table 2.

Table 2 Distribution of Documents and Disease-Drug Associations

Disease Source/Annotation Total Documents Disease-Drug Associations “True” Disease-Drug Associations
Acquired immunodeficiency syndrome RCT/MeSH 270 75 20 (22)
RCT/UMLS 270 106 20
DSUM/UMLS 2003 685 724 35
DSUM/UMLS 2004 805 755 37
Asthma RCT/MeSH 3,349 215 13 (18)
RCT/UMLS 3,349 425 29
DSUM/UMLS 2003 1,332 889 23
DSUM/UMLS 2004 1,457 956 18
Breast neoplasms RCT/MeSH 1,931 191 8 (8)
RCT/UMLS 1,931 210 2
DSUM/UMLS 2003 350 610 4
DSUM/UMLS 2004 391 679 8
Congestive heart failure RCT/MeSH 1,521 246 10 (11)
RCT/UMLS 1,521 433 16
DSUM/UMLS 2003 1,817 1,157 13
DSUM/UMLS 2004 1,916 1,212 22
Diabetes mellitus RCT/MeSH 2,202 172 26 (27)
RCT/UMLS 2,202 241 47
DSUM/UMLS 2003 3,926 874 4
DSUM/UMLS 2004 4,407 894 7
Parkinson’s disease RCT/MeSH 494 80 10 (11)
RCT/UMLS 494 135 5
DSUM/UMLS 2003 211 450 5
DSUM/UMLS 2004 275 525 11
Pneumonia RCT/MeSH 273 116 37 (40)
RCT/UMLS 273 198 105
DSUM/UMLS 2003 1,610 962 31
DSUM/UMLS 2004 1,794 1,036 30
Schizophrenia RCT/MeSH 1,098 186 8 (10)
RCT/UMLS 1,098 241 10
DSUM/UMLS 2003 213 479 23
DSUM/UMLS 2004 232 463 24

This table presents several statistics for the text sources and annotation methods with respect to the eight diseases under investigation. “Total Documents” represents the number of disease-specific documents, “Disease-Drug Associations” refers to the number of 2×2 tables generated for each disease and respective generic name drugs, and “True Disease-Drug Associations” are the number of associations above the identified cutoff used for comparison.

The same set of documents was used for RCT/MeSH and RCT/UMLS.

For comparison, entities in RCT/MeSH represented by MeSH identifiers were mapped to UMLS concepts. Due to one-to-many mappings, two numbers are presented in the “True Disease-Drug Associations” column – one before mapping and one after.