FHIR Chat · extensible binding & preference in alternatives · terminology

Stream: terminology

Topic: extensible binding & preference in alternatives

Patrick Werner (Oct 02 2018 at 15:22):

Is there any mechanism for defining a hierarchical list, or just a favorite, for the alternative ValueSet if you are using an extensible binding?
The closest i can think of is slicing Coding and adding constraints to fill out only one slice at a time. But this wouldn't include the preference for the first sliced coding and would be quite complicated.
Background: In genomics we have a CodeSystem for gene names which everybody agrees to and uses (HGNC), but there are edge cases of genes which haven't a code assigned to (yet). In this case the gene is coded using NCBI coding.

Grahame Grieve (Oct 02 2018 at 16:09):

we do this by slicing

Patrick Werner (Oct 02 2018 at 16:12):

do you have an example? And how do you express the preferred hierarchy between the CodeSystems? E.g.: try CodeSystem_X, if you don't find the code look at CodeSystem_Y, etc...

Richard Townley-O'Neill (Oct 02 2018 at 23:46):

When the element is bound extensibly to a ValueSet, that ValueSet is the primary one, by the definition of extensible bindings.
Any other ValueSets in slices would be of equal priority.

Richard Townley-O'Neill (Oct 02 2018 at 23:48):

Here is something similar used for a profile of STU3 Medication

<element id="Medication.code">
    <path value="Medication.code"/>
    <short value="Coding for the Medicine"/>
    <definition value="Australian coding slices are ..."/>
</element>
<element id="Medication.code.coding">
    <path value="Medication.code.coding"/>
    <slicing>
        <discriminator>
            <type value="value"/>
            <path value="system"/>
        </discriminator>
        <discriminator>
            <type value="value"/>
            <path value="code"/>
        </discriminator>
        <rules value="open"/>
    </slicing>
</element>
<element id="Medication.code.coding:pbs">
    <path value="Medication.code.coding"/>
    <sliceName value="pbs"/>
    <short value="Medication PBS coding (ignore item context)"/>
    <definition value="PBS code from http://pbs.gov.au/code/item. Use of PBS ..."/>
    <binding>
        <strength value="required"/>
        <valueSetReference>
            <reference value="http://hl7.org.au/fhir/ValueSet/pbs-item"/>
        </valueSetReference>
    </binding>
</element>

Simone Heckmann (Oct 02 2018 at 23:59):

So, are you saying, if someone uses a an NCBI code which has an equivalent HGNC code, you want them to use HGNC instead. NCBI is only valid if no equivalent HGNC exists? How is a validator supposed to decide this? Will there be a ConceptMap listing the equivalent pairs of code?

Grahame Grieve (Oct 03 2018 at 11:55):

this is over the horizon for validators at the present

Patrick Werner (Oct 03 2018 at 13:04):

After discussing the issue with @Stefan Lang we came to the solution that you could make a ValueSet including all NCBI Codes + codes from HGNC for all genes that are missing in NCBI. This ValueSet then has to be updated over time to adapt to genes which were included into NCBI in the meantime and therefore the HGNC code will be removed, the NCBI code added for genes new in NCBI

Last updated: Apr 12 2022 at 19:14 UTC