FHIR Chat · ObservationDefinition does not work for the simple stuff · Orders and Observation WG

Stream: Orders and Observation WG

Topic: ObservationDefinition does not work for the simple stuff


view this post on Zulip Eric Haas (Jun 25 2018 at 20:16):

here is my Gist where I prototype the approach that @Grahame Grieve talks about in this chat about using a ObservationDefinition (OD)
to created simple tables ('dictionary') to create a Profile + OD structure instead of creating 100-1000s of profiles using StructureDefinition. I created a spreadsheet that generates Observations from a base profile and I reversed engineered from the spreadsheet to create a OD Logical Model (LM) which I compared to the existing OD

Conclusions:

  • Existing OD doesn't come close to handling this simple yet useful scenario
    * Element names should use the same element names as Observation when they are talking about the same thing
    * The Definition pattern meta data which was missing ( although not used directly in the spreadsheets)

  • Although is not as simple as 1:1 mapping from element to spreadsheet column - My LM comes a lot closer
    * I saved a few elements from it like method and precision:
    * Added the Definition pattern meta data

view this post on Zulip Eric Haas (Jun 26 2018 at 06:37):

Note that each row of a table would be represented by an OD...

view this post on Zulip Hans Buitendijk (Jun 26 2018 at 18:26):

If I get this right, the ObservationDefinition points to a StructureDefinition profile that describes for a "class" of observations (e.g., numeric lab results), the ObservationDefinition attributes that are relevant. Or is the ObservationDefinition using the StructureDefinition to describe for that class of observations what data the Observation is supposed to contain? If the first, then I'm wondering why ObservationDefinition could not be profiled. If the second, we would not be "generating Obserations" rather providing templates of what data to collect for such an Observation. If neither, please clarify.

view this post on Zulip Eric Haas (Jun 27 2018 at 20:40):

The second. OD + Profile = entry in dictionary of Observation templates

you could do the same using a form + Profile where the form is a Questionnaire but then I think you would need to 'decorate' the Questionnaire with a bunch of extensions. @Grahame Grieve ?

view this post on Zulip Grahame Grieve (Jun 28 2018 at 04:06):

umm not sure what you are asking. Dictionary is a technique you can use with several different resources

view this post on Zulip Eric Haas (Jul 14 2018 at 08:07):

@Grahame Grieve can you define what is meant by Dictionary? What I think you mean is that you pull up a particular representation a resource instance which is full of variables that get substituted with the actual values to create an instance. (in my case this is a string representing the json format) This is a shortcut from creation of an SD for each type of measurement.

view this post on Zulip Grahame Grieve (Jul 15 2018 at 04:33):

yes that's pretty much what I mean in practice


Last updated: Apr 12 2022 at 19:14 UTC