13 May 2014

See also: IRC log


DBooth, ericP, +1.480.628.aaaa, +1.202.528.aabb, [IPcaller], Davide, +1.978.794.aacc, Ingeborg, Tony, +1.469.226.aadd


<ericP> scribenick: Claude

I am the scribe

<ericP> ericP: scribing works like this

<ericP> ... blah blah blah

Ingeborg: Currently working as Data Scientist - Working for Booz-Allen Hamilton (?) with disparate data that is standardized. Has an eye towards interoperability for Adverse Events between EHR.

Davide Sottara: From ASU. Area of interest is CDS. Working with Emory Fry and Claude. Interested in clinical data models clinical terminology ontologies and applications in the field.

Thank you

EricP: Focus on FDA Ontologies, HL7 Strategy

FHIR Ontology

<dbooth> Claude: Met with Lloyd McKenzie on FHIR ont, who thought it would be great to include it as part of the DSTU, including RDF as a third serialization. I'll be working w the FHIR team on further dev of the FHIR ont.

<dbooth> ... Now need to tackle profiles, and modifying extensions.

FDA Therapeutic Area ontologies [ericP]

<dbooth> EricP: non-monotonic extensions are challenging

EricP: Therapeutic Ontology Modeling Update: Way to better collect information for semantic trials. Current approaches not so computable.

<ericP> putatively link concept map to webprotege

EricP: Not semantic trials -> Clinical Trials.
... Walking group through diagram components

Address problems they had modeling in the past. E.g., cardiovascular events in two models could not be identified as being coreferent

Thin ontology providing as much interoperability as possible

Currently using actual Clinical Trials data to validate ontology

Provide goal graph for what data looks like so that there is more consistency in the representation of such events.

This provides a graph model and landing pad for definitions even though submission format remains tabular. Terms and semantics would make it into SDTM (?) implementation guidelines. This will require more extensive implementation guidelines. Hope is for Pharma to use BRIDG and represent data in this RDF format rather than using SDTM. Better than shoehorning data into a flat table.

Ingeborg: Makes sense to put this data in RDF format to make it more interoperable and assess outcomes. What are the gaps between where we are now and what we need to get to that point (using BRIDG)

EricP: Shifting implementations to use RDF will probably be less work than defining more complex and specific implementation guides. A leap of faith.
... Interoperability between trials - specific goal of this project.

<ericP> https://github.com/ericprud/FDA-TA/blob/master/tests/RenalTransplantation/scripOutcome.rq

For a particular medication, what was the outcome.

"In general, is this immunosuppressant working

More feasible to 'concatenate' studies.

Big challenge: Reusing BRIDG for all of this. Two problems - fantastically complex. Challenge: Models are often either oversimplified or too complex to lean. How does one find the right middle. Challenge has been learning BRIDG. Not a one-to-one mapping between concept maps and items in the ontology. Folks often don't think in category theory.

HL7 Debrief

David_Booth: HL7 Debrief

Discuss idea of starting workgroup for using RDF for semantic interoperability.

Tony_Malia: Got different track from HL7 meeting - direction refocused after talking with Doug Fridsma

David_Booth: Using RDF at HL7


Initial approach: Representing terminology and FHIR concepts in OWL. Temporarily set aside.

Focus now: Using OWL to represent precoordinated concepts in a post-coordinate fashion.

"Allergy to X"

Allergic Disposition, Substance X

How to represent this in a functional syntax

A paper will result from this.

Represent such post-coordination in a FHIR payload for a CodeableConcept.

<ericP> Claude: would that be part of a terminology system?

Could be done as part of some CTS service.

How does one articulate the system from which this comes?

EricP: Hard part of the existing approach. Precoordination are places where these two models intersect - model is post-coordinate or you use a post-coordinate term with lots of commas, or you use a precoordinated term.

TonyM: What syntax should be used - Functional syntax for expressing precoordination and Manchester syntax. EL should be adequate for this.

EricP: SNOMED uses EL syntax. Approach should work for 99% of this.

Need to announce this to Harold S. from May (did not capture last name)

from Mayo.

<davide_sottara> Harold Solbrig

Thanks Davide

TonyM: Whole group working on this at the VA

Writeup will go from VA to Doug

Probably something to do with common data elements for structured data capture.


<dbooth> Claude: As the ont gets more validated we'll need to go back to the FHIR RDF and align it w the ont. Also need to transform FHIR XML to FHIR RDF and back again. Should discuss this.

<dbooth> ... Still a few areas of the ont that need to be finalized first.

<dbooth> EricP: Look at GenX. That's its purpose.

<dbooth> ... With current FHIR RDF representation we can pretty much round trip. As you adjust per your modeling changes, the script that generates the GenX should be updated. Will be a lot of GenX coding.

<dbooth> ... There's a JavaScript implementation of GenX.

<dbooth> ... Could also produce JSON instead of XML.

<dbooth> ... Python is used to consume XML database to produce GenX.

DavidBooth: HL7 Working Group Update

HL7 work group on RDF for Semantic Interop

<dbooth> http://lists.w3.org/Archives/Public/public-semweb-lifesci/2014May/att-0016/rdf-semantic-interop-wg.doc

Talked to Stan Huff during HL7 meeting about idea of RDF as a universal healthcare exchange language. Using RDF for semantic interoperability. Having things being mappable to RDF. Drafted a charter based on existing one from HL7.

Current version is open for comments

EricP: What are deliverables?

<dbooth> 1. Create and adopt standard mappings from existing and future healthcare information formats, data models, vocabularies and terminologies to RDF and vice versa.

DavidBooth: Create mappings from existing standard so that they can be expressed consistently in RDF

<dbooth> 2. Create and adopt OWL semantic definitions of existing and future healthcare information representation specifications.

Creating OWL ontologies (e.g., FHIR ontology, etc...)

<dbooth> 3. Encourage the use of RDF as a common fallback language for lossless information exchange.

Use RDF as common substrate language to express meaning in messages

<dbooth> 4. Encourage the use of Linked Data principles.

TonyM: Controversial point - Incredible potential for using semantic tools for next generation HL7 artifacts. HL7 has a tendency to create own tooling. Semantic tools could constitute future HL7 tooling platform. Offers a new range and flexibility that was not there in UML. FHIR governance board was reticent about this proposal.

EricP: Lloyd contributed a great deal to tooling due to limitations of UML.

Lloyd was excited with parts of OWL as it addressed some of the shortcomings that the MIF addressed.

Has some buy in on this front.

MIF translated into OWL (O-RIM).

Could these provide an alternative to RoseTree?

<dbooth> Tony: maybe word as "area to be explored"

DavidBooth: Good to address this point but in a non-controversial manner.

TonyM: Introduce the topic in a non-committal way.

Claude: Need to drop off to start another meeting.

Next Meeting

<dbooth> Claude: FHIR ontology topic next week

<dbooth> David: Can also discuss the HL7 work group idea

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.138 (CVS log)
$Date: 2014-05-13 17:25:21 $

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/FDA/Booz-Allen Hamilton/
Found ScribeNick: Claude
Inferring Scribes: Claude
Default Present: DBooth, ericP, +1.480.628.aaaa, +1.202.528.aabb, [IPcaller], Davide, +1.978.794.aacc, Ingeborg, Tony, +1.469.226.aadd
Present: DBooth ericP +1.480.628.aaaa +1.202.528.aabb [IPcaller] Davide +1.978.794.aacc Ingeborg Tony +1.469.226.aadd
Got date from IRC log name: 13 May 2014
Guessing minutes URL: http://www.w3.org/2014/05/13-hcls-minutes.html
People with action items: 

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.

[End of scribe.perl diagnostic output]