W3C

- DRAFT -

Semantic Web Health Care and Life Sciences Interest Group Teleconference

28 Jun 2016

See also: IRC log

Attendees

Present
11AM:_David_Booth, Amol, Shahram, EricP, 5PM:_David_Booth, EricP, hsolbrig, Marc_T, Rob_Hausam, Grahame_Grieve
Regrets
Chair
David Booth
Scribe
dbooth

Contents


ShEx validator

eric: people are using them. There are two, mine and Jose's.
... Harold is using the Fancy ShEx demo interactively to validate, then uses the service to run all examples.

dbooth: Is there a public service available?

eric: There is a service running somewhere, that has harnesses for certain functions. Jose set one up on that.
... There are three implementations of ShEx validators. Harold is working on a fourth.
... Someone at INRIA is working on a fifth.

dbooth: Should we still set up a FHIR shex validator on yosemiteproject.org site?

eric: sure.

dbooth: we should discuss details offline about how to set it up.

Issues List

https://github.com/w3c/hcls-fhir-rdf/issues

https://www.w3.org/2016/06/14-hcls-minutes.html#item03

<ericP> generic shex.js server

Issue 30: What should the type of fhir:link target be in the ShEx?

https://www.w3.org/2016/06/14-hcls-minutes.html#item03

Examples provided by Eric:

# Option 1:

<Obs1> a fhir:Observation;

fhir: Observation.performer [

link <http://hl7.org/fhir/Practitioner/f005>;

Reference.reference [

fhir: value "Practitioner/f005" ] ];

# Option 2:

<Obs1> a fhir:Observation;

fhir: Observation.performer [

link <http://hl7.org/fhir/Practitioner/f005>;

Reference.reference [

fhir: value "Practitioner/f005" ] ];

<http://hl7.org/fhir/Practitioner/f005> a fhir:Practitioner .

# Option 3:

<Obs1> a fhir:Observation;

fhir: Observation.performer [

link <http://hl7.org/fhir/Practitioner/f005>;

Reference.reference [

fhir: value "Practitioner/f005" ] ;

a fhir:PractitionerReference . ];

Here they are again, with typos corrected:

# Option 1:

<Obs1> a fhir:Observation;

fhir: Observation.performer [
... link <http://hl7.org/fhir/Practitioner/f005>;
... Reference.reference [
... value "Practitioner/f005" ] ];

# Option 2:

<Obs1> a fhir:Observation;

fhir: Observation.performer [
... link <http://hl7.org/fhir/Practitioner/f005>;
... Reference.reference [
... value "Practitioner/f005" ] ];

<http://hl7.org/fhir/Practitioner/f005> a fhir:Practitioner .

# Option 3:

<Obs1> a fhir:Observation;

fhir: Observation.performer [
... link <http://hl7.org/fhir/Practitioner/f005>;
... Reference.reference [
... value "Practitioner/f005" ] ;

a fhir:PractitionerReference . ];

MEETING CANCELED -- not enough key people could make it today.

However, the 5pm call will go ahead.

ADJOURNED

========================= 11am ET CALL CANCELED ======================

=================== 5pm Call =======================

FHIR on schema.org

harold: in a proposal that Dr Chute and Dr Chang wrote to the BD2K people, it was suggested to publish metadata elements on schema.org .
... There are a couple of different models for publishing there. One is to publish an entirely separate context.
... Marc has been doing an extension to schema.org , and that's an ontology of components of medicine that would be used to mark up a Mayo site or others.
... We concluded that since we said we would do it, we would publish.
... It had been proposed to use by some of the google people, that there could be value in putting it on schema.org.
... There was also opposition, folks didn't think it would work. But we decided to do a trial baloon, and then open it to the community and evaluate it.
... in order to make it work with the schema.org site, i had to put in schema.org URIs.

<hsolbrig> http://fhir.fhir-schema-org.appspot.com/

harold: One thing we can do is put the related FHIR URIs into the model itself.

<hsolbrig> http://fhir.fhir-schema-org.appspot.com/docs/full.html

harold: If you click "Extension fhir" on that page, and scroll down, you'll see the FHIR organization of the resources.
... Some of the complex datatypes don't work, because i wasn't sure what to do about them.
... There are a number of schema types in schema.org, json, used to mark up HTML pages.
... For example, Cleveland Clinic currently marks up their pages with the info that Marc put together.
... First question: Should this be here at all? Does it have value?
... Renato said that FHIR is for marking up clinical records and they have no business on the web anyway.

Marc: As background, when we started this initiative in 2011 the intention of schema.org was to mark up web resources, so that the big engines can retrieve data.

<hsolbrig> http://my.clevelandclinic.org/health/diseases_conditions/hic_anemia

Marc: in 2012 when we started exchanging electronic health records, when I was working for Agfa, we saw that everyone was using their own resource names.

<hsolbrig> If you view source on the above link, you can see the schema.org medical terms in use

Marc: We said, why not use a public vocabulary? We hijacked the schema.org medical terms, not only in web pages, but to formalize EHR data.

<ericP> SALUS

Marc: When we started doing that it was within SALUS
... At that time, we saw that the vocab was incomplete, and the terms that we wanted to use were not those given.
... So we took an initiative to fine tune that vocabulary, but google and others were very resistant at first. They wanted to see the use case.
... WE showed them, 7 universities were exchanging data this way.

<hsolbrig> http://schema.org/docs/meddocs.html --- this is the work Mark T is talking about

Marc: Using it in a consortium, but eventually exposing the vocabulary on the web.
... Not the private data, but the vocabulary.
... We then got an additional initiative, BioSchemas, in connection with BD2K, and also led by Michel Dumontier at Stanford and others.

<hsolbrig> http://bioschemas.org/

marc: This increased the use of the vocab. But last year we decided with google to first extract the existing vocab, fine tune it, make an extension.
... Next step will be extension, guided by use cases. First is clinicaltrials.gov . They want to publish the trials using the schema. Stanford folks are also involved.
... BioSchema is also suggesting a wide range.
... Health insurance plans also want to use it internally. They have already submitted a patch.

http://pending.schema.org/

http://pending.schema.org/HealthInsurancePlan

marc: The want to use it not only internally, but between health plans in the US.
... When you have a term in schema.org or extension, we put already equivalent class in SNOMED CT.

<ericP> ack ack ericP

trackbot, start meeting

<trackbot> Meeting: Semantic Web Health Care and Life Sciences Interest Group Teleconference

<trackbot> Date: 28 June 2016

eric: The common thread here, in SALUS, the common factor ... schema.org gives us some governance and popularity.
... That is fundamentally in tension with organizations owning their own schema.
... For example, the schema that you get from Spackman's SNOMED ont gives you some URIs.
... But if you put it in schema.org you would have to put a bunch of slashes after it.
... On one hand I'm opposed to a single central schema, because it goes against distributed extensibility of semantic web.
... But schema.org provided an extension mechanism that was not available for the SNOMED ont.

<Marc_Twagirumukiza> +q

eric: Ultimately there is a cost to having a centralization of schemas, that your distributed extensibility atrophies, and you cannot address the long tail
... It does not hurt to have schema.org owning more stuff, but I worry what we're learning about the value of distributed extensibility.

harold: I don't think schema.org is intended to be an ontology publiching org per se.
... But to the extent that they recognize markup in web pages and do intelligent this with them.
... As a primary publishjing mechanism for ont I don't think you should use schema.org.
... But the community can take advantage of some of these big aggregators and write interesting tools.
... But the question Renato raises is whehter we should encourage EHR to be published.

marc: Agree. If you see the way schema was working, my intention from the beginning was to make a branch from schema.org, an extension.
... The core vocab is very hard to put a single term inside. But when you are in an extension it is easy to suggestion something, but the policy of extending terms is that if we do not have a use case we do not add a term.
... To align with vocabs. The intention is to make a kind of hub of vocabs.
... It does not display the SNOMED link, but it is there, and reasoners can use it.
... You can extract all SNOMED equivalent classes.
... The intention is not to create yet another vocab.

grahame: First, i disagree with renato about policy. It is not our business to impose our policy preferences on everyone else.
... If people want to share their healthcare data, they should be allowed to do so.

<ericP> to grahame's point, CDC mines tons of information from tweets

grahame: Regarding fhir on schema.org, I want to understand why the URLs are a problem, and whether it should or sholdn't be the same URLs for identifiers.
... There's a change in perspective between healthcare processes and biomed ontologies. They're veryu diofferent languages. hard to map.
... That seems at the heart of what you would do with schema.org -- exploring those boundaries.
... Use case: medical providers want to hook into patient portal and link from patient healtchare records and knowledge relevant to their conditions.
... info buttons allow them to be linked, but it is not semantic and not clever. There is potential overlap, synergy.

<Zakim> ericP, you wanted to say that health care processes and biomed ontologies connect, but are pretty different use cases on their own

eric: The use cases for writing what is useful care after a heart attack is different from the ont you use to say that "bob had chest pains and our diagnosis was heart attack".

<Marc_Twagirumukiza> +q

eric: They do connect. If you dig into the clinical model, you find out that it is a heart attack. That's the same heart attack that shows up in the biomed ont. But the things that are said about it are different.

grahame: Its the bridging that concerns up most.

marc: Agree. When you look at the use cases, how to formalize healthcare data, but clinical data is different.
... When we started to extend schema.org the intention was to bridge the gap between them.
... For exmaple, how FHIR formalizes medical condition, we can easily use schema.org to put predicates on it.
... Currently not yet done, but in the pipeline, to be published. Started experimenting with equivalentClass.
... Use case is to allow reasoning engine to generate new knowledge.
... Next is to add equivalentProperty.
... Big step is to extend this vocabulary. They are not yet fine enough for clinical use.

harold: There are a couple of additional use cases that came to mind.
... One is the annotation here could be used to mark information that is derived from a FHIR record. Don't know how useful that would be, but FHIR requires an HTML rendering as part of the record. This could add traceability.
... Another place for added value, with personal devices for steps, glucose, heart rate etc., if we had a markup that we could convince the makers to generate, we might get some interesting benefits without necessarily requiring them to use a formal FHIR resource.

marc: I jumped on the work that harold did. Very good info. Mobile devices need things like this. Such use cases would convince these people.
... How to make FHIR on schema.org? First is to discuss if it is useful, worthwhile.
... i am convinced that it is. Need it to converge. How? Need to discuss.
... Could either take the work that harold did already.
... We could do mappings like we did to SNOMED.
... Good to collect information from everyone. Useful?

eric: If we're looking for use cases, the CDC has done projects where they have grepped tweets to get evidence of epidemic.
... if we gave them a nice structured model, they might do better.

dbooth: There was also the well-known google example of predicting the flu before the CDC could.

harold: I owe grahame an explanation and a schema.org expert to tell me where to use schema.org URIs.
... I'll contact marc about that.
... And Dan Brickley. I'll cc folks on that.

<Marc_Twagirumukiza> also post request here: https://www.w3.org/community/schemed/

harold: We can put together a few examples, JSON-LD feeds and other things are kind of interesting.
... Propose in the next week or two to get trial examples put together.
... We're using a github site.

<hsolbrig> https://github.com/crDDI/schemaorg

harold: I'll find the parser to show how I assembled it.

<scribe> ACTION: Harold to put together examples of FHIR on schema.org [recorded in http://www.w3.org/2016/06/28-hcls-minutes.html#action01]

<trackbot> Created ACTION-64 - Put together examples of fhir on schema.org [on Harold Solbrig - due 2016-07-05].

harold: One question: do we need FHIR as it exists now on schema.org? Or should the HCLS references on schema.org include FHIR references?

grahame: We can modify the FHIR publication infrastructure to make it more suitable for that if necessary.

harold: the publication is a simple RDF model. Basically just subclass/superclass-type stuff.

Meeting time

<inserted> dbooth: Tried unsuccessfully to get Renato to join this call. Would like to get him on a teleconference to discuss. Can we schedule a special teleconference to accommodate his timezone need?

eric: Could do one hour later?

grahame: renato mentioned childcare duties at 7am

harold: I'm willing to do 4am or 5am if needed.

dbooth: What about 7am or 8am Eastern US?

grahame: That would be 9pm or 10pm.

<scribe> ACTION: Dbooth to ask renato about scheduling a special teleconference time when he could join, 7am or 8am ET or 6pm ET [recorded in http://www.w3.org/2016/06/28-hcls-minutes.html#action02]

<trackbot> Created ACTION-65 - Ask renato about scheduling a special teleconference time when he could join, 7am or 8am et or 6pm et [on David Booth - due 2016-07-05].

ADJOURNED

Summary of Action Items

[NEW] ACTION: Dbooth to ask renato about scheduling a special teleconference time when he could join, 7am or 8am ET or 6pm ET [recorded in http://www.w3.org/2016/06/28-hcls-minutes.html#action02]
[NEW] ACTION: Harold to put together examples of FHIR on schema.org [recorded in http://www.w3.org/2016/06/28-hcls-minutes.html#action01]
 

Summary of Resolutions

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.144 (CVS log)
$Date: 2016/06/28 22:20:07 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.144  of Date: 2015/11/17 08:39:34  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: s/... Is/dbooth: Is/
Succeeded: s/FHIR validators/validators/
Succeeded: i/Could do one hour later/dbooth: Tried unsuccessfully to get Renato to join this call.  Would like to get him on a teleconference to discuss.  Can we schedule a special teleconference to accommodate his timezone need?
Succeeded: i/5pm Call/Topic: ========================= CALL CANCELED ======================
Succeeded: s/CALL CANCELED/11am ET CALL CANCELED/
No ScribeNick specified.  Guessing ScribeNick: dbooth
Inferring Scribes: dbooth

WARNING: Replacing previous Present list. (Old list: David_Booth, Amol, Shahram, EricP, hsolbrig, Marc, T, Rob_Hausam, Grahame_Grieve)
Use 'Present+ ... ' if you meant to add people without replacing the list,
such as: <dbooth> Present+ 11AM:_David_Booth, Amol, Shahram, EricP, 5PM:_David_Booth, EricP, hsolbrig, Marc_T, Rob_Hausam, Grahame_Grieve

Present: 11AM:_David_Booth Amol Shahram EricP 5PM:_David_Booth EricP hsolbrig Marc_T Rob_Hausam Grahame_Grieve
Found Date: 28 Jun 2016
Guessing minutes URL: http://www.w3.org/2016/06/28-hcls-minutes.html
People with action items: dbooth harold

[End of scribe.perl diagnostic output]