MAKING THE MARRIAGE: MERGING WORLDVIEWS AND MANAGING UPDATES IN MESH AND THE UMLS Stuart J. Nelson, MD, Douglas Johnston, Tammy Powell, and William T. Hole, MD National Library of Medicine, Bethesda, MD

Introduction Faced with two major vocabulary efforts, MeSH and the UMLS Metathesaurus, dealing with significantly overlapping data, the National Library of Medicine (NLM) has begun taking steps towards achieving a common environment in which the vocabularies are developed and maintained. A major step was achieved when the MeSH maintenance environment was redesigned into a modern relational database system, with an important change in the data structure. MeSH moved from a term orientation to an orientation similar to that of the Metathesaurus, that of the concept. Both MeSH and the Metathesaurus carry their own unique identifiers (the MUI and the CUI), presented as the “name that never changes” for each concept, with each maintenance environment carrying the other’s UI as a passive attribute. With each release of the Metathesaurus, it is necessary to reconcile the differences where the previously recorded MUI-CUI correspondence is not still present in the Metathesaurus, as well as to add CUIs to MUIs which have been added to MeSH since the last reconciliation.

Discussion It appears that some reconciliation between the identifiers in the databases will be necessary each year. Because of the differences in the timing and mechanisms of assigning the identifiers, there will always be a need for some reconciliation, or at a minimum, obtaining the new identifiers. However, the differences requiring review will rapidly diminish. Most of the differences resulted from different timing in implementation of changes in definitions of concepts for drugs, publication types, and similarly revised editing policies. Since these revisions took place at a time when the new MeSH maintenance environment was being initialized from the Metathesaurus, the differences in this first reconciliation process appeared more considerable. This process has made us aware of how subtle differences in the maintenance environments or in the operational approach to some deep philosophical issues can affect the vocabulary produced. As in a marriage, sharing of a worldview, a common approach to the operational problems, and a fair and equitable reconciliation process is necessary to bring two separate and distinct entities into unity

The reconciliation of apparent discrepancies between concept structures and previously indicated relationships between unique identifiers is an important step in keeping the Medical Subject Headings (MeSH) and the UMLS Metathesaurus in agreement. This step is a necessary prelude to the planned fusion of the two maintenance environments. A logical analysis of the possible discrepancies allows for recognition of those differences which can be resolved without human review, as well as those which require deeper analysis.