representations of terminologies

Issues

Classes vs. Individuals

A given a hierarchy in some terminology:

 snomed:Infectious-disease-of-lung_disorder
   ⇧
 snomed:Infective-pneumonia_disorder
   ⇧
 snomed:Bacterial-pneumonia_disorder
   ⇧
 snomed:Secondary-bacterial-pneumonia_disorder

can be expressed as a class hierarchy:

 snomed:Infective-pneumonia_disorder rdfs:subClassOf snomed:Infectious-disease-of-lung_disorder .
 snomed:Bacterial-pneumonia_disorder rdfs:subClassOf snomed:Infective-pneumonia_disorder .
 snomed:Secondary-bacterial-pneumonia_disorder rdfs:subClassOf snomed:Bacterial-pneumonia_disorder .

This is appropriate for use cases where the hierarchy is important, such as some CDS system which looks for more than one of the superclasses of snomed:Secondary-bacterial-pneumonia_disorder. If that's not required, one can use a much more efficient closure to group terms into a single-level hierarchy. For instance, if we simply want to collect all forms of infection lung diseases, we can represent the terms as a value set:

 my:Infectious-disease-of-lung_disorder snomed:Infective-pneumonia_disorder owl:oneOf (
 Infectious-disease-of-lung_disorder
 snomed:Bacterial-pneumonia_disorder
 snomed:Secondary-bacterial-pneumonia_disorder ).

This is more efficient, at the expense of expressivity (the hierarchy is gone). If we know the instance data will only use e.g. leaf-level codes, we can even eliminate the intermediate terns:

 my:Infectious-disease-of-lung_disorder owl:oneOf (
 snomed:Secondary-bacterial-pneumonia_disorder ). # plus viral + ...

Concept Descriptor vs. IRI

Most clinical record systems use something like an ISO 21090 CD element with attributes to uniquely identifying a concept.

The most critical these are the pair of codeSystem and code. These effectively form a unique identier for a concept. The concatonation of these can be viewed as a URN (like a URL, but not dereferencable). Using the HL7 O-RIM Coding for 21090, the attributes look like:

 :labObs5678 # Joe's baseline measured in body fluid
   a renal:RheumatoidFactorObservation ;
   hl7:coding [ dt:CDCoding.code "13930-3" ; dt:CDCoding.codeSystem "2.16.840.1.113883.6.1" ; # or EVS or ...
                dt:CDCoding.displayName 
   "Rheumatoid factor:Dilution Factor (Titer):Point in time:Synovial fluid (Joint fluid):Quantitative:Agglutination" ;
                dt:CDCoding.codeSystemName "LOINC" ] .

Viewed as a URN, this could look like:

 :labObs5678 # Joe's baseline measured in body fluid
   a renal:RheumatoidFactorObservation ;
   hl7:coding <urn:oidplus:2.16.840.1.113883.6.1/13930-3> .

with some helpful metadata:

 <urn:oidplus:2.16.840.1.113883.6.1/13930-3>
   dt:CDCoding.code "13930-3" ; dt:CDCoding.codeSystem "2.16.840.1.113883.6.1" ; # or EVS or ...
   dt:CDCoding.displayName 
   "Rheumatoid factor:Dilution Factor (Titer):Point in time:Synovial fluid (Joint fluid):Quantitative:Agglutination" ;
   dt:CDCoding.codeSystemName "LOINC" .

It is easy to use OWL equivalent class to assert the equivalence between these two forms:

 <urn:oidplus:2.16.840.1.113883.6.1/13930-3> 
   owl:equivalentClass [
     owl:intersectionOf (
       [ owl:onProperty dt:CDCoding.codeSystem ; owl:hasValue "2.16.840.1.113883.6.1" ]
       [ owl:onProperty dt:CDCoding.code       ; owl:hasValue "13930-3" ]
       ) ] .

Resources

example of LOINC as classes
alpha prototype of LOINC in OWL format.
LOINC Ontology (under development) captures axes.
Another LOINC Ontology captures the complete set of LOINC codes.
Semantic Web Representation of LOINC: an Ontological Perspective pubmed/AMIA.
NCI CD page