The only missing part, now that I look at the SNOMED Compositional Grammar <> and Expression Constraint Language <> specs, is how to create a URI (which is the type of a term binding in ADL2 <>) from a post-coordinated expression or constraint expression. This should be trivial, but I don't see where SNOMED has specified it.

- thomas

On 03/05/2017 11:23, Thomas Beale wrote:

Hi Koray,

On 03/05/2017 10:41, Koray Atalag wrote:

Hi again,

I’ve been snowed under for a while and just now catching up with this…I reckon there was a suggestion that we do not include SNOMED codes within archetypes, or more specifically post-coordinated expressions, if I understand correctly but to define these somewhere else and then include the external URI instead. While this would be a good solution for well-defined expressions, subsets etc. I think if you think about the vast amount of potential expressions with almost endless permutations of terms it quickly becomes too complex and unmanageable. Therefore there will always need for including specific expressions within archetypes and templates.

I’ve been doing a lot of terminology bindings using various ontologies and terminology lately and I think we need urgently a consistent way to make these bindings and get the tools support it. For example when term bindings (for the purpose of defining real-world meaning of a node) are done at archetype level you end up with local at codes that refer to each binding and then it is possible to link one or more terms from same or different terminology systems. For the purpose of providing a valueset to a DV_CODED_TEXT at archetype level we don’t have a very clear way – we keep on saying we’ll put a terminology query but it is not really usable or useful.

I'm unclear on the problem exactly. We can bind to a URI that points to a SNOMED CT subset. If we upgrade the binding syntax in ADL 1.4 as per this SPECPR-215 <>, then we can include SNOMED constraint expressions.
Let's assume we do this, and the tools follow suit.
What else is needed? (I agree that we need to address all this ASAP by the way. It is already addressed in ADL2, but the user-base is moving much more slowly than I expected towards it).

Tooling support is also not satisfactory. But when you do that at template level (e.g. define a valueset for a DV_TEXT by further constraining it to DV_CODED_TEXT or an existing DV_CODED_TEXT) it is just a list of terms, code and terminology system with version/release. It is not clear how we can refer to an external list defined by a terminology query or refset – at least I couldn’t figure out. There’s quite an inconsistency between archetype vs template defined valuesets within .opt – whereas they should be defined in the same way and share same semantics.

I don’t know how to fix it – my guess is that this is not a commonly used feature so it was never a high priority for SEC group. I think it is time to bring loose ends together as more and more countries adopt SNOMED and there is clear pressure to do this. FHIR terminology service is quite good and I think we should just start using it. If we need further bells and whistles it can be extended.

I'm not sure how using the FHIR terminology service solves the above problems. As far as I can see, the question of FHIR or CTS2 or xyz terminology service is orthogonal to the question of representing bindings to post-coordinated expressions.

- thomas

openEHR-technical mailing list

Reply via email to