W3C

- DRAFT -

HCLS

10 Jun 2014

See also: IRC log

Attendees

Present
David_Booth, Mehmet, Kerstin_Forsberg, Tony_Mallia, Neda, EricP, Claude, Ingeborg, EmoryFry, Marc_Belgium
Regrets
Chair
EricP
Scribe
ericP, dbooth

Contents


<inserted> ScribeNick: ericP

<Marc_Belgium> This is Marc

<Marc_Belgium> I failed to start Zakim

<dbooth> marc, will you be joining the teleconference?

<Marc_Belgium> am having problem to join

<dbooth> anything i can help with?

<Marc_Belgium> yeah ... all I need is to call the bridge +1.617.761.6200 right?

<dbooth> marc, yes and then the code is 4257

Draft HL7 Charter

<dbooth> http://lists.w3.org/Archives/Public/public-semweb-lifesci/2014Jun/att-0014/rdf-semantic-interop-wg-v2.doc

-> http://lists.w3.org/Archives/Public/public-semweb-lifesci/2014Jun/att-0014/rdf-semantic-interop-wg-v2.doc hl7 draft charter

<dbooth> https://join.me/523-323-048

<Kerstin> works fine

<Marc_Belgium> +1 got it

<dbooth> Eric: Maybe change "encourage the use of" to "support the use of"

<dbooth> Eric: Act as expert resources to help say how RDF and OWL can be used

ericP: s/encourage/support # act as expert resources

<dbooth> EricP: Support other working groups in both OWL modeling adn RDF data expression.

<dbooth> https://join.me/523-323-048

s/sorking groups/Working Groups (e.g. RCRIM, FHIR)/

<dbooth> EricP: Could also propose a working group for only RDF-izing FHIR

<Claude> David, you mention interoperability and lossless representation but I would also stress that RDF will also support the computability and inferencing of clinical information which is a key goal of Clinical Decision Support and Clinical Quality Information

ericP: we could scale down to FHIR

<Marc_Belgium> IMHO We should not limit on RDF-izing FHIR...

dbooth: i see this as wider

<dbooth> Tony: What's the basis of future healthcare systems? This is even bigger than just the exchange of data.

Tony: we want to model health care in general

Claude: how do we capture data that used to be unstructured in a structured way?

<Marc_Belgium> exact the task is wider and 'Semantic interoperability ' Layer on HL7 standards means all

<dbooth> EricP: i think HL7 would think of themselves are providing enough to enable structured data for everything.

Claude: providing structured data that is interoperable across different HL7 standards
... e.g. a notation of criticality used in FHIR and CDS

ericP: they will be suspicious that we will do better with RDF than they did with CDE

Claude: i can define my FHIR model using 4 different, local-specific terms for criticality, but all bound to one concept

<Marc_Belgium> +1 this is an expected suspicious situation -- but not a problem

Tony: IHTSDO already has ontologies in OWL

<dbooth> Tony: those two efforts should be referenced

Tony: we need to reach out to them

<dbooth> ... ICD-11 is the other

<Marc_Belgium> ICD-11 is not yet there

Tony: the integration between structural models and terminology models has not gone well
... note that HL7 has not quality standards for how interop is tested

<Marc_Belgium> they have such plan but I think we should not list them here unless we have a clear communication with them and know where they are

<Marc_Belgium> There are several other external efforts

<dbooth> David: Question: should this work group define concepts or develop ontologies?

<Marc_Belgium> like SMART Platform

<Kerstin> URI standard for SNOMED CT - http://ihtsdo.org/fileadmin/user_upload/doc/download/doc_UriStandard_Current-en-US_INT_20140527.pdf

Tony: people have to make arbitrary decisions and then they run into collisions

<Claude> Is this group a subgroup of the TSC?

<Claude> Oops, not TSC

dbooth: our work should NOT be defining common concepts

<Claude> I forgot the acronym

<Marc_Belgium> I would not support developping ontologies here...

Tony: <sarcasm>the world needs yet another health care model</sarcasm>

<Marc_Belgium> but use existing ones

Tony: should this group focus on the convergence
... so what is the work product?

<Claude> The ITS Work Group supports the HL7 mission through the development of Implementable Technology Specifications and Messaging Protocol guidelines (e.g. ebXML, Web Services, Messaging encoding specifications) to be used when implementing the HL7 information specifications.

dbooth: specific goals [read from charter]
... create RDF views of existing data

<Marc_Belgium> +1

Tony: this is ontology mapping

dbooth: if RDF weren't adopted as a representation of FHIR, we could define a mapping from FHIR-JSON to RDF

Tony: that's A-box

dbooth: T-box is the next bullet

Tony: this 1st is instances, 2nd is models
... so the first bullet read like it was about T-box

Marc_Belgium: we have several efforts: SMART Platforms, FHIR-RDF, ...
... what we need to produce is a pattern graph which can be mapped to e.g. FHIR et. al
... we need an abstract model that can be used for other models. this would help interop

<dbooth> David: That's what I meant by common data element.

Claude, ericP, we don't want to invoke CDE if it wasn't considered successful

scribe: but with RDF, we can define a vague notion

<Claude> Claude is raising his hand

dbooth: reponding to suggestion about defining CDEs:
... in essence, that's what every WG tries to do
... i think it's important to work at a higher level
... instead of guessing the common models, work on the process that enables sharing

<Zakim> dbooth, you wanted to say that any effort for a common data model or common data elements is in essence the same thing that every specific domain standard tries to do.

Claude: i think that design-by-constrain, class-based (vs. set-based) historically fails

<dbooth> claude: the reason this was not successful previously is that classes were defined and it was a closed approach. It was class based rather than set based.

Claude: the tech that hl7 was using for semantic bridge is ill-suited compared to RDF/OWL

dbooth: i can see benefits in that regard, but i've seen it many tines before

<dbooth> xkcd/927

<Marc_Belgium> leavinf for another meeting

<dbooth> thank you marc!

<Marc_Belgium> will keep following this great initiative

<eafry> eric and David….I have a comment

<dbooth> Claude: HL7 has ITS group. They looked at XML representation and made comments. If we bring in RDF, we should thing about how it relates to the ITS group.

Claude: when we were involved in @@1, the ITS group provided [serialization] advice
... we should look hard at the relationship to ITS

emory: most of the discussion this AM has perhaps inadvertently been about data model
... the vocabulary needs to evolve as well.

<dbooth> Emory: Most of the discussion this morning has been about data models. Suggest we enforce or stress that semantic vocabularies need to evolve also, so that we can do subsumption and other reasoning. There's an effort in ICD-10 and SNOMED, but there may be a role to support vocabulary involvement also.

emory: the current charter doesn't emphasize the fact that vocabularies are as important as structural model

<Mehmet> https://join.me/523-323-048

<dbooth> EricP: would like to see examination of use cases where different parties interpreted standards differently, and figuring out exactly what is needed for interop. Create a crystal of interop and build upon it.

<dbooth> David: Those use cases are extremely valuable to work through.

<dbooth> EricP: Give people as few choices as possible except around UI.

<dbooth> David: Ideal common data element is only ideal for one party, not for another with a different use case.

<dbooth> EricP: But then you have to look at the cost of conversion if they are different.

<dbooth> Claude: Need clear use cases -- makes it concreted.

Tony: there's a critical point beyond which one can't unify and one must instead translate

dbooth: i don't want to fall into the trap of thinking we can come up with the right CDEs
... hard to predict; might emerge over the next 30 years

Tony: i'd like to see differentiation between transformation and unification
... i think RDF and OWL are good for exploring the differences and connecting

Claude: we can eliminate this artificial barrier between the model and the terminology

dbooth: so "uniformly relate models and terminology"?

-> http://www.w3.org/wiki/HCLS/ClinicalObservationsInteroperability/TermInfo

-> https://www.timetag.tv/wric/play/23453 terminfo at 17m00s ?

<scribe> scribenick: ericP

<scribe> scribenick: dbooth

<ericP> dbooth, i can't remember if rrsagent handles n scribenicks

<ericP> you can emperically test this by re-drafting mins and see if they changed

<ericP> (cause i just added you as a scribenick)

<scribe> ScribeNick: ericP

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.138 (CVS log)
$Date: 2014-06-10 17:57:27 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.138  of Date: 2013-04-25 13:59:11  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/sorking/working/
Succeeded: s/ontologies/modeling/
FAILED: s/sorking groups/Working Groups (e.g. RCRIM, FHIR)/
Succeeded: i/This is Marc/ScribeNick: ericP
Found ScribeNick: ericP
Found ScribeNick: ericP
Found ScribeNick: dbooth
Found ScribeNick: ericP
WARNING: No scribe lines found matching ScribeNick pattern: <ericP> ...
Inferring Scribes: ericP, dbooth
Scribes: ericP, dbooth
ScribeNicks: ericP, dbooth
Default Present: DBooth, Mehmet, Kerstin_Forsberg, +1.978.794.aaaa, Tony, +1.469.226.aabb, Neda, ericP, Claude, +1.301.825.aacc, Ingeborg, EmoryFry, Jos_De_Roo
Present: David_Booth Mehmet Kerstin_Forsberg Tony_Mallia Neda EricP Claude Ingeborg EmoryFry Marc_Belgium
Got date from IRC log name: 10 Jun 2014
Guessing minutes URL: http://www.w3.org/2014/06/10-hcls-minutes.html
People with action items: 

[End of scribe.perl diagnostic output]