Quantifying the Impact and Extent of Undocumented Biomedical Synonymy
Figure 1
Very little information is shared across multiple biomedical terminologies.
(A) The panel on the left illustrates the overlap among the concepts annotated by the terminologies documenting Diseases and Syndromes. The figure itself is composed of ten concentric rings, with the outermost ring (k = 1) indicating the colors assigned to each dataset. The next ring (k = 2) displays the overlap in concepts among all pairwise comparisons, arranged in clockwise order starting with the intersection (MSH, NCI). The extent in overlap was computed by dividing the number of co-occurring annotations by the maximum possible number given the sizes of the terminologies being intersected (percent maximum overlap). This information is displayed within the concentric ring using bi-colored bars, whose heights depict the percent maximum overlap for the terminologies indicated by the colors. The panels on the right illustrate this idea by enlarging a section of the original figure, highlighting a particular intersection (NCI, CHV), and explaining how the colored bar translates into the percent maximum overlap. The remaining concentric rings (k = 3…10) display the overlap extent for all higher order intersections (3-way, 4-way, etc.), with each ring containing colored bars. (B) This figure illustrates the overlap among terms annotated to each concept for the same ten datasets depicted in (A). (C, D) These panels show the overlap in concepts (C) and terms (D) for the Pharmacological Substances terminologies. Note that only the ten largest datasets were included in each panel for the sake of clarity.