In This Issue:

• Home
  
  
• Director's CorNER
  
  
• Ill Advised
  
  
• Inside Outreach
  
  
• PubMed Particulars
  
  
• Tres Chic
  
  
• Santa and Digital Delivery
  
  
• Sister Library
  
  
• Announcements
  
  
• Links
  
  

OK, so it's not usually like that in the health sciences. Here, it's more like this: you say "acetaminophen," I say "Tylenol," and that guy over there says "paracetamol." In a general sense, we're all talking about the same thing. That's the naming problem: lots of different terms that name the same meaning.

Back in 1985, Dr. Lindberg testified before the House Committee on Appropriations and proposed a long-term research and development program he felt was needed. He said, in part:

What Dr. Lindberg and others at NLM foresaw was that the naming problem, as complex as it was in 1985, was going to get worse. Not only would each biomedical database use its own terminology, but each new computerized component of the health system would add to the growing number of ways to name the same meaning. Computerized billing records would develop coding systems. Electronic patient records would develop systematic terminology. Radiologist would devise a scheme to encode their information. Each hospital and every medical school would develop terminology and coding systems. Everyone would get in on the act.

So, the naming problem would get exponentially worse.

More importantly, Dr. Lindberg and NLM realized that in order for all these different computerized pieces to communicate effectively there had to be a way to solve that naming problem.

Oh, many of you are saying, that's why we have controlled vocabularies. That's right. Every controlled vocabulary used for indexing and/or cataloging is an attempt at dealing with the naming problem. And, in one sense, it's an effective solution. Look at our favorite example, MeSH. It's been an incredibly powerful tool. But, now think about the fact that there is MeSH and there is the CINAHL vocabulary and there is the PsycINFO vocabulary and there is the College of American Pathologists' SNOMED (Systematized Nomenclature of Medicine) vocabulary and the ICD-10 and HCPCS and the CPT and the hundreds of other vocabularies out there.

So, I say myocardial infarction. You say 410 [the ICD-9 code]. And your aunt the RN says MI. The guy in the waiting room says heart attack. And while we may be able to make that mental leap to understand that we're all talking about essentially the same thing, computers can't unless they know all those different coding/vocabulary systems.

And that's the point of UMLS:

Currently, UMLS takes over 90 different vocabularies and coding systems and not only finds and maps the synonyms to each other, but describes the semantic relationships between words in order that the computer system can literally "understand" all those vocabularies. That's a very simplistic description of the complex process of integrated vocabularies into UMLS, but, that's the idea.

The concept is NOT to build the uber-vocabulary. UMLS preserves each of the individual vocabularies inviolate. It is what UMLS adds that's important - and that is the understanding of the relationships between each of these.

UMLS will "enable computer systems to understand biomedical meaning in user questions and to use this understanding to retrieve and integrate relevant machine-readable information for the user." [UMLS class, Oct. 2002, Willis & Tilley]

So, you're thinking, why is this part of the PubMed particulars column? Ever wonder why the MeSH browser in PubMed will never map "heart attack" to "myocardial infarction"? But, if you type "heart attack" into the search box in PubMed, it WILL map to "myocardial infarction"? That's the power of UMLS at work behind the scenes in the Entrez system.

If you use the "Find Terms" button in the NLM Gateway, you'll also see UMLS at work. Try it sometimes. Just look for some terms. And, make sure you look for some medical terms and some non-medical ones. Try "pizza" just for the fun of it.

UMLS is improving searching and retrieval in bibliographic databases such as PubMed and the Gateway. Its development has been the result of the very hard work of a lot of very bright people at NLM and elsewhere. I've certainly only discussed the most simple concepts involved in UMLS here. I encourage us all, as librarians and searchers, to learn more about UMLS. You can do that by looking at: http://www.nlm.nih.gov/research/umls/

Donna Berryman, Outreach Coordinator