text and description

Technical (archive)
21

Hi!

I know that there are suggestions for defining terminology
queries/subset-selections using URIs. We discussed this a bit in a
conference paper that was selected for republication in BMC:
"Integration of tools for binding archetypes to SNOMED CT" freely
available at http://www.biomedcentral.com/1472-6947/8/S1/S7

This kind of URI-encoded queries with semantics have come and gone and
seem to be coming again in openEHR discussions. Sometimes the URIs
have contained semantics similar to what you describe below. Sometimes
they have just contained ID's of queries that have their semantics
hidden, sorry I mean stored/maintained, in some query server. See
especially the appendix in the paper above for discussion and
references.

However, my recent question/suggestion did not have much to do with
those URI-encoded terminology queries.

Instead, I was asking about two specific use-cases:
1. directly pointing out single codes/concepts that already have URIs and
2. a way of creating "local" bindings using URIs as unique identifiers.

Please re-read the previous post and feel free to ask more if I have
not made the difference clear enough.

Best regards,
Erik Sundvall
erisu@imt.liu.se http://www.imt.liu.se/~erisu/ Tel: +46-13-227579

22

Erik,
I don't see the difference. The same approach can be used with different
query parameters and terminology identifiers. We just need to find a way to
uniquely identify local terminology ids, some sort of namespacing mechanism
such as a terminology source organisation UID should do the trick. This
would not be that much different to uniquely identifying templates which is
also under development.

Heath

From: e.sundvall@gmail.com [mailto:e.sundvall@gmail.com] On Behalf Of Erik
Sundvall
Sent: Tuesday, 2 December 2008 7:03 PM
To: For openEHR technical discussions
Cc: Heath Frankel; hugh.grady@oceaninformatics.com
Subject: Re: text and description

Hi!

I know that there are suggestions for defining terminology
queries/subset-selections using URIs. We discussed this a bit in a
conference paper that was selected for republication in BMC:
"Integration of tools for binding archetypes to SNOMED CT" freely
available at http://www.biomedcentral.com/1472-6947/8/S1/S7

This kind of URI-encoded queries with semantics have come and gone and
seem to be coming again in openEHR discussions. Sometimes the URIs
have contained semantics similar to what you describe below. Sometimes
they have just contained ID's of queries that have their semantics
hidden, sorry I mean stored/maintained, in some query server. See
especially the appendix in the paper above for discussion and
references.

However, my recent question/suggestion did not have much to do with
those URI-encoded terminology queries.

Instead, I was asking about two specific use-cases:
1. directly pointing out single codes/concepts that already have URIs and
2. a way of creating "local" bindings using URIs as unique identifiers.

Please re-read the previous post and feel free to ask more if I have
not made the difference clear enough.

Best regards,
Erik Sundvall
erisu@imt.liu.se http://www.imt.liu.se/~erisu/ Tel: +46-13-227579

> Hi Erik,
> As you know Ocean has been doing a lot of work making terminology and
> openEHR Archetype work. Hugh Grady is the best to describe this but in
> summary we are proposing the use of terminology URIs for bindings.
>
> Bindings can reference a whole terminology, a branch of a terminology
> hierarchy or a complex query which extracts specific subset of a
> terminology.
>
> To identify these there at least four identifiers; terminology ID,

subset

> ID, query name and query version id. There are other parameters such as
> language and terminology version.
>
> In simply cases where you just want to reference a terminology it might

look

> something like the following
> (NOTE: these are examples to illustrate the point and are certainly not

a

> final proposal).
> terminology:snomed-ct?language=en-GB
>
> or for a specific version of SNOMED
> terminology:snomed-ct(2003)?language=en-GB
>
> For a hierarchy of a terminology it might look something like
> terminology:snomed-ct(2003)/hierarchy?rootConcept=28374832
>
> and for a pre-specified query
> terminology:snomed-ct(2003)/query?name=AllBacteria
>
> There are also more specific URIs for terminology queries by using

subset

> and query version identifiers (UIDs) mentioned above.
>
> I believe this work is ongoing and is being proposed through IHTSDO. I
> suggest we wait for that work to conclude but I thought I would let you

know

> that Erik's thinking is certainly the way things are being proposed.
>
> Heath
>
>> From: openehr-technical-bounces@openehr.org [mailto:openehr-technical-
>> bounces@openehr.org] On Behalf Of Erik Sundvall
>> Sent: Monday, 1 December 2008 11:20 PM
>> To: For openEHR technical discussions
>> Subject: Re: text and description
>>
>> Hi!
>>
>> Would it be a good or bad idea to have URI:: as a valid terminology
>> prefix in openEHR terminology bindings, with the intention to host...
>>
>> 1. "local" bindings that are not foreseen to be of public general use:
>>

URI::http://www.cs.chalmers.se/~oloft/terminologies/odont-123/local-Mucos-

>> txtur
>>
>> 2. Potentially universally interesting terminologies that already have
>> official URIs but do not (yet?) have openEHR-defined prefix:
>> URI::urn:miriam:obo.go:GO%3A0045202
>>
>> I guess opening up for any URIs would lead to a risk of having double
>> representations (URI+openEHR-prefix) for the same thing, like...
>> URI::urn:UMLS/CID=C0037658
>>
>> ...and the example URI::urn:miriam:obo.go:GO%3A0045202 is just one of
>> several URI-ways to point out an entry in the gene ontology..
>>
>> What are the other pitfalls and/or benefits?
>>
>> I guess there will probably never be only one ultimate updated
>> registry fitting every purpose, not from openEHR, not from EuroRec not
>> from anybody else.
>>
>> Best regards,
>> Erik Sundvall
>> erisu@imt.liu.se http://www.imt.liu.se/~erisu/ Tel: +46-13-227579
>>
>> P.s. Remember that URIs include both URLs and URNs
>>
>> > Dear all,
>> > The European Institute for Health Records has created a registry of
> coding
>> > systems.
>> > In the (near) future they expect to be the place where coding systems
> and
>> > their meta-information are registered so an URL and unique

identifying

>> > number will suffice.
>> > Will this be the way to go?
>> > Gerard
>> >
>> >
>> > -- <private> --
>> > Gerard Freriks, MD
>> > Huigsloterdijk 378
>> > 2158 LR Buitenkaag
>> > The Netherlands
>> > T: +31 252544896
>> > M: +31 620347088
>> > E: gfrer@luna.nl
>> >
>> > Those who would give up essential Liberty, to purchase a little
> temporary
>> > Safety, deserve neither Liberty nor Safety. Benjamin Franklin 11 Nov
> 1755
>> >
>> >
>> >
>> >
>> >
>> > So custom/local terminologies can be handled this way and the
> implementation
>> > will be left to developers....BUT this may result in different
>> > implementations which may render interoperability in the long run....
>> >
>> > So I suggest a sub-section within ontology section where used
> terminologies
>> > are declared explicitly; i.e. "umls": 2008AA version of NLM UMLS
> knowledge
>> > sources. Perhaps an URI and other details can be specified (i.e.

WSDL).

> I
>> > think it is easier for the community to agree on such a naming
> convention.
>> >
>> > Custom local terminologies can be declared this way and you can

create

>> > terminology names for use in term/constraint bindings.Perhaps

creating a

23

Hi,

Some days after 2008-12-09 IHTSDO probably release a draft of a
compositional grammar for SNOMED CT expressions which can be used for
terminology bindings to SNOMED CT. This draft can probably be a basis for
discussions of terminology bindings.

  Greetings,
  Mikael

24

Hi Thomas,

According to Support_IM 5.3.2

"Valid identifiers that can be used for this attribute for terminologies include but are not limited to the
following:
• “openehr”
• “centc251”
• an identifier value from the first column of the US National Library or Medicine (NLM)
UMLS terminology identifiers table below, in either of two forms:

  • as is, e.g. “ICD10AM_2000”, “ICPC93”;
  • with any trailing section starting with an underscore removed, e.g. “ICD10AM”."

So, as far as I can see it should be possible to use a local identifier, although not supported by the editors at present, the issue being how to namespace the terminology uniquely.

I could not see anything in other documentation
For Olof’s project I am proposing to use e.g.

“se.chalmers.MUKOS::mukos-reaktion”

In real use, and if we have designed the archetypes optimally, most of these term-bindings should be at template-level rather than archetypes.

Dr Ian McNicoll
office / fax +44(0)141 560 4657
mobile +44 (0)775 209 7859
skype ianmcnicoll
ian@mcmi.co.uk

Clinical Analyst - Ocean Informatics ian.mcnicoll@oceaninformatics.com

Member of BCS Primary Health Care Specialist Group – www.phcsg.org

2008/12/1 Thomas Beale <thomas.beale@oceaninformatics.com>

25

Thomas,

I do not have the details, but I know they use the CEN standard for registries.

Francois Mennerat will know more.

Gerard

– –
Gerard Freriks, MD
Huigsloterdijk 378
2158 LR Buitenkaag
The Netherlands

T: +31 252544896
M: +31 620347088
E: gfrer@luna.nl

Those who would give up essential Liberty, to purchase a little temporary
Safety, deserve neither Liberty nor Safety. Benjamin Franklin 11 Nov 1755

26

Ian McNicoll wrote:

Hi Thomas,

According to Support_IM 5.3.2

"Valid identifiers that can be used for this attribute for
terminologies include but are not limited to the
following:
• "openehr"
• "centc251"
• an identifier value from the first column of the US National Library
or Medicine (NLM)
UMLS terminology identifiers table below, in either of two forms:
- as is, e.g. "ICD10AM_2000", "ICPC93";
- with any trailing section starting with an underscore removed, e.g.
"ICD10AM"."

So, as far as I can see it should be possible to use a local
identifier, although not supported by the editors at present, the
issue being how to namespace the terminology uniquely.

I could not see anything in other documentation
For Olof's project I am proposing to use e.g.

"se.chalmers.MUKOS::mukos-reaktion"

Ian,

only problem - that identifier would not be allowed in the (imminent)
Release 1.0.2 of openEHR (see
http://www.openehr.org/svn/specification/BRANCHES/Release-1.0.2-candidate/publishing/roadmap.html)

4.3.12.1 Identifier Syntax
The syntax of the value attribute is as follows:
# -------- production rules --------
terminology_id: name [ ‘(’ version ‘)’ ]
name: V_NAME
version: V_VERSION
# -------- lexical patterns --------
V_NAME: [a-zA-Z][a-zA-Z0-9_-/+]+
V_VERSION: [a-zA-Z0-9][a-zA-Z0-9_-/.]+

27

Same patten less the dots should be ok then?

e.g. sechalmersMUKOS::

Ian

Dr Ian McNicoll
office / fax +44(0)141 560 4657
mobile +44 (0)775 209 7859
skype ianmcnicoll
ian@mcmi.co.uk

Clinical Analyst - Ocean Informatics ian.mcnicoll@oceaninformatics.com

Member of BCS Primary Health Care Specialist Group – www.phcsg.org

2008/12/2 Thomas Beale <thomas.beale@oceaninformatics.com>

28

Hi Mikael

I've not seen the final specification, but in it's original
draft, it was extremely limited in scope. For instance, the
language would have made the grammar unsuitable for use in
OpenEHR, though there was no technical reason for this.
Apparently there's going to be a new draft shortly.

However even in it's revised form, the document will almost
certainly only specify a syntax for building single concepts.
It will not provide a syntax or a grammar that deals with
conformance and binding, though one of these is surely needed.

Grahame

Mikael Nyström wrote:

29

Hi All

I think we are addressing an issue that will come up in templates as well. How to add terms from a local terminology directly into a template. Local is used for the archetype terms – I have wondered if we should use a template: namespace for terms that are created only in the template but for internationalisation.

Cheers, Sam

30

This was the exact approach that was proposed in work within Standards Australia to represent Archetyped data in HL7 v2. The at-codes defined in the ontology section of the archetype was used in a coded element concept ID and the archetype ID was used as the coding system (compared with “local” when refer to internally). This approach could certainly be extrapolated to localised coded terms specified in templates using the template ID (assuming that it is globally unique).

Heath

31

Hi Grahame,

You are right that the compositional grammar don’t solve the whole problem.
I just meant that the compositional grammar probably is a part of the
solution for binging to SNOMED CT (and maybe other systems that is possible
to postcoordinate.)

  Greetings,
  Mikael

32

The problem with that is that it’s just an identifier, containing no information.
Whether one would use sechalmersMUKOS or chalmersMukosSe or chSTw23r17 would make no difference.

The suggestion from Erik about using URI:s seems a bit more appealing to me.

regards

Olof

2 dec 2008 kl. 19.14 skrev Ian McNicoll:

33

Sam Heard wrote:

Hi All

I think we are addressing an issue that will come up in templates as
well. How to add terms from a local terminology directly into a
template. Local is used for the archetype terms – I have wondered if
we should use a template: namespace for terms that are created only in
the template but for internationalisation.

I would expect to do it in the same way as for archetypes - binidings.
Just the same, but in templates.

- thomas beale