W3C HCLS – General Meeting -- 06 Nov 2012

scribenick bobP

Michel: First thing, acknowledge that group composition is changing
... over last year was experiment: three areas, two task forces for each
... but onerous and confusing (a little)
... So...whether to restructure
... Matthias would like to merge Pgx and CDS
... suggesting to go to three task forces: LS, PGx, health care

<matthias_samwald> Sounds very good, won't impact participation.

ericP: COI has been happening outside of HCLS, it's me and Charlie

<egombocz> I think, it's a good suggestion - it will increase participation in my opinion as it condenses the number of meetings, and there is some logical / thematic overlap

ericP: CDISC work may bring in more pharmas to to clinical trials discussion for calls
... otherwise, we can have more conversations, spawn calls when we need them

Scott: +1 to fewer meetings; there was interest in joining HCLS, simpler to merge, plus participation will go up
... always open, but not perceived that way: Wiki page should be Step1, step 2, step 3
... if you call in at wrong time discussion will seem opaque

ericP: Not sure we get more participation with fewer meetings
... need action items and public humiliation :D
... items and tracking and meeting due dates
... also, Task force leads do most of the work; we need fresh blood, motivated to find people who get our problems

Michel: Identifying what the tasks are; and how to encourage people to take up tasks
... always have talked about getting funding, but w/in our own research group for our own tasks
... for participant: What do they want to get out of it?
... drive conversation thru level of commitment, or go for more resources?
... alignment of activity and participant is the issue

ericP: +1 to action items at particular times

Michel: Actionable action items

Rich: Attracted to calls b/c drug interations, linking w clinical relevance etc
... +1 to restructuring according to use cases
... there is overlap between what HCLS does and my goals
... value is collaborating with people

Michel: HCLS has fostered new collaborations

Rich: It has been useful to have artifacts

Michel: Agree. Major role of linked life data group, extending our hand to other communities
... a natural role for w3c and hcls to get up-take in other communities

<egonw> I found that a lodd group was easier to work for/with, as the tasks were simpler, and very explicit

<egonw> use cases are much harder to solve, and practically converted into actions points

Ratnesh: Specific task leaders, at the end they have a recommendation; who owns the task?

<egonw> the action points tend to take a lot more time

ericP: Reason these are productive is that the chair rules with an iron fist.
... comes back to Michel's point if ID-ing action items
... paid professional time for WG, but here people work on the margins of their time
... how much you can expect of somebody
... tasks that are near HCLS could turn into work w/in HCLS
... Matthias: how is CDS is going?

Matthias: Participants in calls are short of time
... many of us, problem is lack of time plus lack of clear goals
... prototype of RDF is reasonable goals 5 years ago; not reasonable now

<ratnesh> +1 to Matthias, HCLS needs clear cut goals

Matthias: need goals that have impact. Harder to organize in heterogeneous group w/o hierarchy
... CDS, I'm working on release of ontology that captures PGx rules, alleles

<matthias_samwald> http://www.w3.org/wiki/HCLSIG/CDS/Example_of_individual_pharmacogenetic_patient_data_in_OWL

Matthias: (this shows PGx for a single person?)
... difficult to divide up this task. Prototype will be soon, then establish this ontology based formalism

ericP: So, awkward time right before clean-up, check.

Michel: Lot of work has been preliminary; engage point is when you can show something
... lot of stuff we build gets noticed, but after the fact
... discussion about prototype comes before buy-in

ericP: +Prototypes then show them off
... OWLim stuff is proceeding mostly on its own
... CDISC, working on what ODM looks like in RDF, must spend time on getting value out
... term invocation, how to find equivalences in clincially acquired data
... hoping to get more pharmas in CDISC stuff
... I type a lot :)

Scott: CDISC to clinical, or HL7, FHIR?

ericP; There is blog about how to do this in xml. Working now on how to do this clearly in RDF

scribe: do not need to make a choice between cdisc or other, etc

Scott: Take advantage of several participants in group. Ratnesh is part of link for safety
... Charlie was enthusiastic: take RDF for eligibility for trial, use that w patient data

ericP: Sounds like it would be good for COI front page
... are there partners that can work w us?
... I'm spread thin right now
... but therer are revolutions going in both HL7 and FDA

Ratnesh: I can write about clincial trial stuff

ericP: Another piece: whole bunch of IMIs that do overlapping work

<mscottm> My proposal more precisely was: Take the RDF/OWL encoding of eligibility criteria for clinical trials (formal representation to

<mscottm> come from VUA) and use it to steer info retrieval from a partner's patient data (initially exported as HL7 v2/v3 XML).

ericP: we could help community adopt semweb "tech w least frustration" document the intersection points

<mscottm> The goal is to bridge from RDF to HL7 models with eligibility studies as the 'driving' application.

<ericP> http://www.w3.org/wiki/HCLS/SWAT4LS2012/Hackathon#Data_Map

ericP: there are modeling efforts before sparql endpoints; we can help people make better use of their time
... "Hey, let's become librarians"

Michel: Hackathon list of communities is great. Exactly the approach to be a bridge between communities

ericP: If bridges are critical enough then the value persists
... we can show practical ways to take advantage of our mapping efforts
... doc that is crucial that engages community to maintain

<ratnesh> http://www.w3.org/wiki/HCLS/ClinicalObservationsInteroperability/HL7_RIM

ericP: We can figure new ways to describe what we have done
... breakdown by dataspace, by task, different axes for people to comprehend where new people should enter
... can improve our outreach greatly

Ratnesh: Link gives an overview of HL7 models
... RIM, everything is UML diagrams. But for ontology, some things are not obvious

ericP: Lloyd MacKenzie(?) is also modeling the RIM in OWL
... HL7 RIM, xml is a MIF, MIF to OWL; Lloyd is working on all this

Ratnesh: (discussion of classes, attributes, identities etc)

(Ratnesh, ericP, Michel go way-way deep :)

ericP: +1 appreciation to Ratnesh who speaks both OWL and UML

<ericP> http://www.w3.org/wiki/HCLS/ClinicalObservationsInteroperability/CDISC

ericP: CDISC, groups Roche and AZ are working on SDTM

<egombocz> Apologies -need to log out for another meeting - excellent meeting today!

ericP: Combining clincial care data and clincial trial data, no distributed extensibility model
... will write this down in RDF

<michel> sorry gotta run!

Scott: Patients/trials: we need not everything about pt, but start by eligibility criteria
... narrows down from thousands of things to say to tens
... criteria from trial is the place to start, instead of swimming in data

ericP: Nothing that disambiguates w/in EHR data

<ratnesh> OWLRIM: OWL-UML differences http://www.w3.org/TR/sw-oosd-primer/

Scott: Eligibility criteria is about the Pt, so it's the same core domain as the EHR

ericP: But the same domains are being captured in different ways
... yes the intersect captures the person and the protocol, but it is hard to see how to use it

Scott: Encode eligibility requirements...

ericP: Clinical trial criteria, each application reduces the aggregate
... CROs get precise data about people, sometimes you want to look at later in 2nd study
... also subfractions of cohorts
... working toward useability once the trial has already starte

- DRAFT -

W3C HCLS – General Meeting

06 Nov 2012

Attendees

Contents

Summary of Action Items

Scribe.perl diagnostic output