Standardized clinical vocabularies are just as essential as standardized phenotypic data models. The lack of widely accepted standardized vocabularies for clinical care has greatly hampered the development of automated decision support tools and clinical research databases. The impossibility of guaranteeing that a serum sodium or systolic blood pressure has the same code or term throughout our hospital system is troublesome. Fortunately, several efforts in the private and public realm, e.g., Logical Observation Identifiers, Names, and Codes (LOINC) , are addressing this issue. The National Library of Medicine has invested large resources in the Unified Medical Language System (UMLS) to enable these different vocabularies to be interoperable, at least at a basic level [120, 158]. Most of the most widely used vocabularies, such as Reed codes , Medical Subject Headings (mesh) , and Systematized Nomenclature of Medicine (SNOMED)  are represented in the UMLS.
The same problems are not unknown in bioinformatics as is pointed out in the section on nomenclature (section 5.5). Many microarray technologies use their own system of accession numbers, which then have to be translated into a more widely used nomenclature such as LocusLink.
As is the case for data models, as functional genomic investigations venture into the domain of clinical correlates of gene sequence, expression, and protein concentrations, the lack of standardized clinical vocabularies will prevent large-scale accumulation of useful data at the national or even regional level. For example, if a set of expressed genes are thought to be associated with a particular set of cardiac arrhythmias, this association will only be found if the nomenclature used to describe the arrhythmias in the clinical databases is drawn from one of the standardized vocabularies. Additionally, these vocabularies would have to be used with the same semantics at each data collection site. Rather than inventing yet another competing standard for a clinical vocabulary, bioinformaticians should examine which of the existing component vocabularies within the UMLS best suits the needs of a particular set of studies.
Was this article helpful?