W3C

- DRAFT -

Semantic Web Health Care and Life Sciences Interest Group Teleconference

17 Feb 2015

See also: IRC log

Attendees

Present
Tony_Mallia, Brian_Pech, Paul_Knapp, Kerstin_Forsberg, Peter_Hendler, Guoqian_Jiang, Rob_Hausam, David_Booth, Marc_Twagirumukiza, Claude_Nanjo, Lloyd_McKenzie, Dale_Nelson, Sajjad_Hussain
Regrets
Chair
David Booth and Paul Knapp
Scribe
dbooth

Contents


<trackbot> Date: 17 February 2015

<Sajjad_Hussain> [GVoice] is Sajjad_Hussain

<Marc_Twagirumukiza> Zakim ??P7 is Marc

<Marc_Twagirumukiza> +Marc_Twagirumukiza

<Sajjad_Hussain> [GVoice] is Sajjad_Hussain

gotomeeting info: https://global.gotomeeting.com/join/157514853

Access Code: 157-514-853

Approval of minutes

Feb 3 minutes: http://wiki.hl7.org/index.php?title=ITS_RDF_Concall_Minutes_20150203

RESOLUTION: Feb 3 minutes unanimously

Feb 10: http://wiki.hl7.org/index.php?title=ITS_RDF_Concall_Minutes_20150210

RESOLUTION: Feb 10 minutes unanimously approved

FHIR Ontology Requirements -- Formally approve

http://wiki.hl7.org/index.php?title=FHIR_Ontology_Requirements

RESOLUTION: FHIROntology requirements unanimously approved!

Common model

http://wiki.hl7.org/index.php?title=RDF_for_Semantic_Interoperability#Common_model_for_data_element_management_and_value_set_binding

Guoqian: Proposed to use ISO 11179 for common model elements.
... I put together a wiki page, would like your input on the scope.
... Want to focus on 11179 as core model. Also look into reference models by CIMI community and CTS2 value set spec.
... What do you think? Useful that we should work on in this group?

Tony: Surrounding question: We need to start to scope the projects in this group accurately, to avoid wandering.
... We had been focusing on FHIR RDF exchange. Does that include 11179 focus? Maybe we're starting to have a number of different projects. Should start to define what the projects are.
... How to look at the agenda for the mtg? Purely FHIR?

Peter: To do more meaninful or deeper analysis, we'll need entities and roles in acts. CIMI has an alternative ref model, not as good for ref. Should not add another model. Same answer as last week: not part of this project.

Lloyd: Nervous about using 11179 as base model for bindings. It doesn't have extensible valueset bindings, and no notion of SHOULD or MAY instead of SHALL. Both of those will make life entertaining in RDF, but we can support it.
... Also we have a resource in FHIR called DataElement that is intended to support what's in 11179, so it may be that in figuring out how to do bindings in FHIR we could also support bindings from 11179 perspective.

<Marc_Twagirumukiza> +1 for idea of defining what the projects to work on are (practical organization of activities) and may be this can pop up

Lloyd: Not a fan of doing this through CTS2, because it's evil.
... Merging would be hard.

David: This work group is broader than just FHIR.

Guoqian: If we only focus on FHIR you would be right, but this group is broader.
... We have a requirement to harmonize data elements from other models, so 11179 nmight be a good candidate harmonizingn different models.

David: This group is required by its charter to be data model and vocabulary agnostic.

claude: There are cases where we'll want to link HL7 work with other work. Might make sense for this group to have an important role in that.

lloyd: What about limitations I mentioned -- extensible valueset bindings, and no notion of SHOULD or MAY?

Guoqian: 11179 has a good model for the valuesets, but not sure it fits well into FHIR.
... But based on our use cases here, use 11179 for metadata about data elements, then use CTS2 services for valuesets.

Lloyd: Not specific to FHIR, but the concern is for HL7 in general.
... How to make it work if the ont is based on 11179?

Guoqian: We have some use cases at Mayo. Harold Solbrig is familiar with 11179 and CTS2 models.
... Ask Harold to join a future call to provide input?

Answers: Claude: yes. Dale: yes. Kerstin: yes.

<scribe> ACTION: David to ask Harold if he can join a future call to discuss 11179 and CTS2 [recorded in http://www.w3.org/2015/02/17-hcls-minutes.html#action01]

<trackbot> 'David' is an ambiguous username. Please try a different identifier, such as family name or username (e.g., dderour, dhansen2, dhau2, dnewman, dshotton, dwild).

<scribe> ACTION: David to put Lloyd on agenda to discuss valuesets in FHIR [recorded in http://www.w3.org/2015/02/17-hcls-minutes.html#action02]

<trackbot> 'David' is an ambiguous username. Please try a different identifier, such as family name or username (e.g., dderour, dhansen2, dhau2, dnewman, dshotton, dwild).

Guoqian: Maybe in a future conference call we can present work at Mayo on data element harmonization

Terminology binding in FHIR RDF

<Marc_Twagirumukiza> we see it

(Tony presents live demo in Protege)

http://lists.w3.org/Archives/Public/public-semweb-lifesci/2015Jan/0019.html

Tony: I tried representing an instance as a class, but that seemed too revolutionary.
... Coding as element, with two forms: Code, System and Display.
... Coding would be the linkage point for terminologies.
... Substance is subclass of Resource.
... I'm in an ont that has individuals, and it refers back to FHIR model, different ont is imported, though protege shows them all.
... Record 01339 is a substance and has a substance type.
... Started looking at the mappings -- how to do uniquness of the properties?
... I used dot notation: Substance.type points to a Coding (instance is 1349).
... Display code in the system are still object properties.
... Code is an instance, and it has a type of FHIR Code.
... Has a code value. Type is a FHIR code primitive.
... Started to run into other interpretatinos of the mapping: TQ conversion of XML to OWL didn't quite do this, but primitives in FHIR would map to datatypes in RDF.
... Code primitive is XSD token with length 1.
... Data primitive will look like this. Code primitive would be a datatype, and we've got some consistency.
... We can get the structural value of the coding instance, and it maps to the values.
... Also able to apply another type to the Coding -- a semantic type.
... Structural type was Coding, but semantic type is URI pointing to the SNOMED code inside SNOMED ont.
... Raises quite a few questions. How to bind instance to types during transform?
... How do you know it is a substance FHIR resource?
... Binding happens to be the same name as the type, but not sure it is reliable.
... Leave it to the reasoner to deduce the bindings? Put it on the Range of the objectProperty? Always bind to that type, so that the reasoner can figure out the type?
... Returning to the coding, we have an instance of coding, declared it type FHIRCoding, and it has structural values, and complex types have values carrying the base data.
... Coding is basically the same as how FHIR uses it.

Lloyd: Looks good. We can't declare in the instance the type of the coding. We've got one ont w SNOMED IDs, and a separate FHIR ont with properties of codings. Need a third ont that goes through SNOMED IDS, that connects coding values to types.
... Need to do that for whatever ont we want to link in for validation.
... Strategy makes sense.
... Also, there may be multiple coding systems, via codable concept.
... Means that instead of setting range for coding, we'll say that one of the codings must meet the valueset constraints, but it doesn't matter which one.

Tony: transformations may need to be computational -- not just ont.

<Marc_Twagirumukiza> In snomed a concept can have multiple parent in hierarchy. If you declare an instance being element of a snomed class what will be the relationship between this instance and the parents of this class?

lloyd: Challenge: URLs for IDs will vary between code systems -- not deterministic.
... Requires knowledge of the coding system.

Guoqian: You want to use SNOMED code to indicate the meaning of FHIR substance?

Tony: No, FHIR substance type: House Dust.
... Substance.type property said what substance it was.

Guoqian: You put SNOMED code here to indicate the type of link element?

Tony: Yes, this is FHIR sample.
... you're asking: Does there need to be a binding between the type substance and the SNOMED class for the substance? That would not be indicated for the instance, but between the types.
... Mappings class-class.
... The instance is being mapped to the type -- both to its structural type, but also to its SNOMED code directly.
... Instance has a type (or types), but the range is defined by the class above it, that says what the valueset range can be.
... In SNOMED ont type substance is generic, but it goes down the hierarchy

<Marc_Twagirumukiza> OK! got it

Marc: When you map an instance to a class, but the class is in a hierarchy, how do you deal with the hierarchy?

Tony: I don't deal with them, but the classes in SNOMED have multiple parents. I just need to bind it at the right level.
... Need to get to some heavy duty lifting at this level of detail.
... Maybe a subgroup to work on this?

claude: Agree. Need to figure out how to structure this weekly meeting, but FHIR ont will require a lot of work. Inevitable that we'll need a separate mtg for that.

Dale: Subgroup notion would be welcome. Procedurally, alignment with existing subgroup would fit the notion of a project.

Tony: Need project scope statement?

Dale: Yes, then passes it up the chain to ensure that all interested parties are informed and can join and avoid duplication of effort.

ADJOURNED

Summary of Action Items

[NEW] ACTION: David to ask Harold if he can join a future call to discuss 11179 and CTS2 [recorded in http://www.w3.org/2015/02/17-hcls-minutes.html#action01]
[NEW] ACTION: David to put Lloyd on agenda to discuss valuesets in FHIR [recorded in http://www.w3.org/2015/02/17-hcls-minutes.html#action02]
 
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.140 (CVS log)
$Date: 2015-02-17 21:06:37 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.140  of Date: 2014-11-06 18:16:30  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/Type/type/
Succeeded: s/tual/tural/
Succeeded: s/ Ontoloogy/Ontology/
Succeeded: s/unaimously/unanimously/
Succeeded: s/2q+//
No ScribeNick specified.  Guessing ScribeNick: dbooth
Inferring Scribes: dbooth
Default Present: Tony, bpech, pknapp, Kerstin_Forsberg, peterhendler, Guoqian, rhausam, DBooth, +1.916.783.aaaa, Marc, +1.310.266.aabb, claude, Lloyd, DNelson, Sajjad_Hussain, [IPcaller]
Present: Tony_Mallia Brian_Pech Paul_Knapp Kerstin_Forsberg Peter_Hendler Guoqian_Jiang Rob_Hausam David_Booth Marc_Twagirumukiza Claude_Nanjo Lloyd_McKenzie Dale_Nelson Sajjad_Hussain
Found Date: 17 Feb 2015
Guessing minutes URL: http://www.w3.org/2015/02/17-hcls-minutes.html
People with action items: david

[End of scribe.perl diagnostic output]