The practical impact of ontologies on biomedical informatics
To examine the development of ontologies and the structures of knowledge contained within.
Review published literature using PubMed and full-text searches into recent Medinfo and American Medical Informatics (AMIA) Symposia proceedings, using "ontology" and "ontologies" as search terms.
This article reviews the following set of controlled terminologies
Galen, Unified Medical Language System (UMLS), Medical Entities Dictionary (MED), SNOMED-CT, LOINC, Foundation Model of Anatomy (FMA), Gene Ontology, ISO Reference Terminology for Nursing Diagnosis, NDF-RT, RxNorm, NCI Thesaurus, and DOLCE+. These terminologies were chosen based recentness and impact, or potential to make an impact, in their respective fields.
Galen was designed to represent terms independent the recording language (in English with added terms for disambiguation) and any health record its encoded on. UMLS is an amalgamation of terminologies. Terms are broken down into three parts ==
semantic term, concepts, and relations between concepts. One or more semantic terms make up a concept and two or more concepts make up a relation. MED uses terminologies from various departmental systems and applications. SNOMED-CT employs Description Logic (DL) to have a collection of phrases define a concept. LOINC used a structured naming system to be applied in a laboratory setting. FMA was used to provide a fine grained approach to describing anatomic terms; the terms are specified in a hierarchy of body parts.
GO is a controlled biological terminology made up of three parts: biological process, cellular components, and molecular functions. ISO Reference Terminology is a collection of nursing interventions, referred to as the "Loose Canon". Terms are modeled after nursing judgments, containing categories such as focus, judgment, potentiality, etc. NDF-RT and RxNorm is part of a push to standardize drug terminologies. The NCI Thesaurus was designed to support cancer research terminologies. DOLCE+ is a high level, domain independent, conceptual framework used for disambiguation.
The developments in ontologies indicate a push towards having a structured framework for knowledge. Each of these terminologies developed to an insufficient or lack of proper tools available to encapsulate or disambiguate concepts within each domain. Their impact is apparent through their frequent use in the literature and in practice throughout the Biomedical fields. The article goes on to cite specific sources in which they are used for research, laboratory results, health records, and surgical procedures. The use of a standard makes it easier to communicate specific meaning between systems and people. As a result, their increased adoption may yield improvements in healthcare.
The next big challenge, which the author touches on briefly, would be to develop a system to translate between terminologies. Of course, this would be a difficult task at best, considering the inherent structural differences between each ontology. The fact that there is no One ontology, is indicative of how each terminology has evolved to define a specific domain effectively.