- From: Forsberg, Kerstin L <Kerstin.L.Forsberg@astrazeneca.com>
- Date: Tue, 6 Nov 2007 11:08:53 +0100
- To: "Alan Ruttenberg" <alanruttenberg@gmail.com>
- Cc: "Vipul Kashyap" <VKASHYAP1@PARTNERS.ORG>, <public-hcls-dse@w3.org>, "Stanley Huff" <Stan.Huff@intermountainmail.org>, "Oniki, Tom (GE Healthcare, consultant)" <Tom.Oniki@ge.com>, "Jonathan Rees" <jar@creativecommons.org>
Thanks Alan, so, if you use NCIt's concepts in the concept structure for Organ it would be fine to use their conceptid:s, and assigned URI:s, to denote for example the liver as a "thing" or "type of entity"? What concerns me is the way CDISC pick terms from many different concepts structures in NCIt, combine them into controlled terminologies, and use them as strings (submission values) in the Topic fields in SDTM records. (see http://www.w3.org/2007/11/NCIt_and_CDISC_CT.ppt) In your article about record/statement/domain level, http://owl-workshop.man.ac.uk/acceptedLong/submission_26.pdf, I think you call to this as the "database / record attitude". What I try to understand is how the way forward that you describe in the article for BioPAX database records can be applied for clinical observations so that we enable secondary use and ensure "meaningful automated inference"? - Kerstin On Nov 4, 2007, at 4:20 PM, Forsberg, Kerstin L wrote: >> [AR] I note that SDTM has places where one uses NCI Thesaurus, >> which is, for a large part, a domain model. > > Kerstin: What part of NCI Thesaurus would you say is a domain > model? My understanding is that Barry would refer to it as a > "concept system" and not as a domain model / ontology of real world > of entities/phenomenon 1) There are a couple of papers of Barry's specifically about the NCIt at [1]. The second paper notes flaws, but places at least a portion of the ontology as a domain ontology. The distinction between conceptualists and realists in the paper you cite is the difference between whether you have the point of view that the classes in your ontology are whatever sets of thing (real or not) that you want them to be, versus whether the classes are sets [*] of particular things in the real world that are the same sort of thing. So, in NCIt, the hierarchy below "Organ" seems to me to be trying to be a domain ontology of organs. OTOH, Look at terms below Training and Education. The subclass "Fellow of American College of Physicians", would be a concept. Dosage_Form_Not_Applicable, with definition "The use of a dosage form term is not relevant or appropriate" would seem to be part of the record level, describing the fact that it doesn't make sense to put some data in some field. -Alan [*] Barry will complain of my use of "classes are sets" here. It is, strictly speaking, wrong, but makes the point easier to see. [1] http://ontology.buffalo.edu/medo/NCIT_Smith.html > > The way I see that CDISC have used picked concepts from different > concept hierarchies in NCI Thesaurus to construct controlled > terminologies are also quite confusing. See for example the > examples I list in my previously posted slides http://www.w3.org/ > 2007/11/NCIt_and_CDISC_CT.ppt. > > > 1) http://ontology.buffalo.edu/concepts/ConceptsandOntologies.pdf > > > Regards > Kerstin
Received on Tuesday, 6 November 2007 10:09:12 UTC