the meaning of the words used in a patient medical report is the first step to understanding what is wrong with the patient. The National Library of Medicine's (NLM) Unified Medical Language System® (UMLS®) has created several resources to "facilitate the development of computer systems that behave as if they 'understand' the meaning of the language of biomed-icine and health'' (www.nlm.nih.gov/research/umls/about_ umls.html). The NLM freely distributes three UMLS knowledge sources: the Metathesaurus®, the Semantic Network, and the SPECIALIST Lexicon. The three knowledge sources can assist NLP applications in understanding the meaning of the words in clinical reports.
The Metathesaurus is a vocabulary database of biomedical and health related concepts containing over 900,000 concepts compiled from more than 60 different source vocabularies. The Metathesaurus integrates existing vocabularies (such as SnoMed and ICD-9), which provide terms and sometimes hierarchies relating the terms. The Metathesaurus organizes the terms into concepts, organizes the concepts into hierarchies, and relates concepts to each other. If a concept from a new source vocabulary already exists in the Metathesaurus, the concept is added as a synonym. The Metathesaurus is the most complete collection of biomedical concepts and their synonyms.
The Semantic Network provides a consistent categorization of all concepts represented in the Metathesaurus, which are the nodes in the network, and provides a useful set of relations among these concepts, which are the arcs in the network. Every concept in the Metathesaurus is assigned at least one of 135 different semantic types (e.g., finding, anatomical structure, pathologic function, etc.). The Semantic Network contains 54 relationships among the semantic types, such as "part of,'' "is-a,'' and "caused by.''
An NLP application for our SARS detector may find the phrase "shortness of breath'' in a patient report, which is a synonym for the Metathesaurus concept Dyspnea. Other synonyms for the concept Dyspnea are "difficulty breathing,'' "SOB,'' and "breathlessness.'' The concept Dyspnea has the semantic type of Sign or Symptom and has children like "hypoventilation,'' "paroxysmal dyspnea,'' and "respiratory insufficiency.'' A knowledge base with synonyms and semantic information can be helpful in identifying variables and their values from text.
The SPECIALIST Lexicon is a general English lexicon that includes many of the biomedical terms in the Metathesaurus together with the most commonly occurring English words. As of 2003, the SPECIALIST contained almost 300,000 entries. A lexical entry for each word or term records information about spelling variants, derivation, inflection, and syntax. Using the SPECIALIST, we could know, for example, that the term "mediastinal'' is the adjectival form of the noun "mediastinum'' and that the following phrases are equivalent: "mediastinal widening,'' "widened mediastinum,'' and "wide mediastinum.''
Was this article helpful?