- From: Lloyd McKenzie <lloyd@lmckenzie.com>
- Date: Wed, 26 Nov 2014 08:42:16 +0100
- To: David Booth <david@dbooth.org>
- Cc: w3c semweb HCLS <public-semweb-lifesci@w3.org>, "its@lists.hl7.org" <its@lists.hl7.org>
- Message-ID: <CAJ860JL_eSa_VV+7az2YuewakgGnrBLqFM4yy281v+16T05nAw@mail.gmail.com>
Sorry I missed the call (and I'm likely to miss the next one too). The general representation of FHIR in RDF/OWL is pretty straight-forward. FHIR has Resources which are structures that can contain sub-structures. The structures correspond to classes. The elements within those structures correspond to properties. However, there are several tricky bits we need to deal with: 1. All elements are extensible, even the simple types. So birthDate isn't actually a simple type, but instead a complex type with an optional "value" property with type xsd:date, but can have other properties in addition/instead 2. In addition to "regular" extensions, resources can have modifierExtensions which change the meaning of at least some other properties. That sort of behavior gives reasoners fits, but we need to account for it 3. The terminologies we link to are continuously evolving. FHIR allows identifying the version of the code system a code is from - do we have to make that matter? I did it in the RIM ontology, but it's painful. 4. Coded elements can have multiple codings and the meaning of the codings won't necessarily be equivalent. The solution will have to work in an environment where multiple codings are declared and mappings exist between the codes 5. In FHIR there are defaults and information can sometimes be conveyed by the absence of information. E.g. If a patient doesn't declare the "animal" element, it's inferred to be a human. This doesn't play that nicely with an open-world view where failing to declare "animal" doesn't mean that patient isn't an animal. 6. In FHIR, repeating elements are treated as lists/arrays where order matters and must be retained. 7. Finally, the RDF sytntax needs to be round-trippable with the XML and JSON syntaxes. I.e. An instance must be convertable from RDF to JSON to XML to RDF without any loss of information and with the conversion process not knowing anything other than the FHIR resource and data type definitions. (And ideally, the round-tripped version should still be valid against any digital signatures.) The RDF (FHIR instance) and OWL (FHIR profile definition) representations will need to take into account the above. (It's possible I've missed a couple of "gotchas" too - so feel free to supplement this list.) Lloyd -------------------------------------- Lloyd McKenzie +1-780-993-9501 Note: Unless explicitly stated otherwise, the opinions and positions expressed in this e-mail do not necessarily reflect those of my clients nor those of the organizations with whom I hold governance positions. On Tue, Nov 25, 2014 at 7:09 PM, David Booth <david@dbooth.org> wrote: > Draft minutes from today's call: > http://www.w3.org/2014/11/25-hcls-minutes.html > and also below in plain text. Thanks to Tony for an excellent intro to > his FHIR ontology approach! Tony's slides: > http://lists.w3.org/Archives/Public/www-archive/2014Nov/ > att-0048/FHIR_OWLv3.pdf > > Next week will will continue with EricP's and Claude's FHIR ontology > approaches. > > Thanks! > David Booth > > -------------------------------------- > > [1]W3C > > [1] http://www.w3.org/ > > - DRAFT - > > HCLS with Tony Mallia on FHIR ontology > > 25 Nov 2014 > > See also: [2]IRC log > > [2] http://www.w3.org/2014/11/25-hcls-irc > > Attendees > > Present > Tony_Mallia, David_Booth, Josh_Mandel, Rob_Hausam, > EricP_Joshua_Phillips, Patricia_Grime(sp?), Stan_Huff, > Cecil_Lynch, mscottm, Guoqian, Charlie_Mead, > Marc_Twagirumukiza, Rafael_Richards_(IRC_only?), > Daniel_Karlsson, Cati_Martinez_Costa_(IRC_only?) > > Regrets > Chair > DavidBooth > > Scribe > ericP > > Contents > > * [3]Topics > 1. [4]Role call and agenda > 2. [5]Logistics > 3. [6]Orphaned action items > 4. [7]Review of Work Projects > 5. [8]PhUSE-FDA project (formerly CDISC2RDF) > 6. [9]C-CDA RDF representations > 7. [10]High-level concept mapping to RDF (AR typeCodes, > etc.) > 8. [11]Comparison of FHIR ontology approaches > * [12]Summary of Action Items > __________________________________________________________ > > <JoshM> So should I be dialing into gotomeeting, or the W3C > bridge? > > <dbooth> zakim aadd is PatriciaGrime > > <Claude_> 1. Please join my meeting. > [13]https://global.gotomeeting.com/join/157514853 2. Use your > microphone and speakers (VoIP) - a headset is recommended. Or, > call in using your telephone. Dial +1 (213) 289-0016 Access > Code: 157-514-853 Audio PIN: Shown after joining the meeting > Meeting ID: 157-514-853 GoToMeeting® Online Meetings Made Easy® > > [13] https://global.gotomeeting.com/join/157514853 > > Role call and agenda > > <inserted> Postponed approving last week's minutes, because no > ITS co-chair is on the call to permit formal approval. > > <daniel> daniel karlsson here, not identified > > Logistics > > <scribe> ACTION: ericP to set up tracker [recorded in > [14]http://www.w3.org/2014/11/25-hcls-minutes.html#action01] > > <dbooth> Orphaned action items > > Orphaned action items > > Review of Work Projects > > <dbooth> ACTION: Tony to find out more details about how iCat > handles ICD-11 ont and report back [recorded in > [15]http://www.w3.org/2014/11/18-hcls-minutes.html#action01] -- > PENDING > > [15] http://www.w3.org/2014/11/18-hcls-minutes.html#action01] > > PhUSE-FDA project (formerly CDISC2RDF) > > <dbooth> ACTION: Kerstin and Ingeborg to prepare a status and > future state ideas for PhUSE-FDA work [recorded in > [16]http://www.w3.org/2014/11/18-hcls-minutes.html#action05] -- > PENDING > > [16] http://www.w3.org/2014/11/18-hcls-minutes.html#action05] > > C-CDA RDF representations > > <dbooth> ACTION: Eric to establish/make a wiki page for C-CDA > RDF representations work [recorded in > [17]http://www.w3.org/2014/11/18-hcls-minutes.html#action06] > > [17] http://www.w3.org/2014/11/18-hcls-minutes.html#action06] > > <dbooth> [PENDING] > > High-level concept mapping to RDF (AR typeCodes, etc.) > > <dbooth> ACTION: Tony and Rob to report their plan on > High-level concept mapping to RDF work [recorded in > [18]http://www.w3.org/2014/11/18-hcls-minutes.html#action08] -- > PENDING > > [18] http://www.w3.org/2014/11/18-hcls-minutes.html#action08] > > <dbooth> ACTION: Rob and all to decide on a wiki for Term Info > work [recorded in > [19]http://www.w3.org/2014/11/18-hcls-minutes.html#action09] -- > PENDING > > [19] http://www.w3.org/2014/11/18-hcls-minutes.html#action09] > > <dbooth> Should be an action for Rob instead of Tony. > > -> > [20]https://www.w3.org/wiki/HCLS/ClinicalObservationsInteropera > bility/TermInfo HCLS TermInfo page > > [20] https://www.w3.org/wiki/HCLS/ClinicalObservationsInteropera > bility/TermInfo > > Comparison of FHIR ontology approaches > > <dbooth> gotomeeting link: > [21]https://global.gotomeeting.com/join/157514853 > > [21] https://global.gotomeeting.com/join/157514853 > > <scribe> scribenick: ericP > > Tony: this is exploratory work to see the issues around an > ontology for FHIR > ... w3c defines many exchange formats > ... RDFXML is widely used > ... will present in Turtle 'cause ericP proposed that. i never > look at the bits so RDF/XML is fine for me > ... in FHIR, you exchange a fragment of an ontology > ... but if you're using a SemWeb environment, you can import > other ontologies to see the full picture > ... the refs to resources are like FHIR resources > > <dbooth> Tony Mallia's FHIR ontology slides: > [22]http://lists.w3.org/Archives/Public/www-archive/2014Nov/att > -0048/FHIR_OWLv3.pdf > > [22] http://lists.w3.org/Archives/Public/www-archive/2014Nov/ > att-0048/FHIR_OWLv3.pdf > > Tony: but in terminologies, you express it as an RDF URI which > you have gotten from some other mechanism > ... in the exchange, we see that it's a system of ontologies: > ... .. exchange > ... .. terminology > ... .. instance record > ... . > ... next step: work out the distributed ontology > ... fhir types point to terms. you change what is mostly the > facts to a recieving system which maps it to pictures > ... slide 4: ballot adverse reaction example > ... example symptoms is embedded rather than a resource > ... exposure has an external link to the substance > > dbooth: is this an appropriate example? > > cecil: by "embedded objects", you mean attributes of the > resource itself > > Tony: we have a mixture of patterns which we need to deal with > ... , need to figure out how we'd map from one to another > ... [slide 6] > ... an XML element name points to a type in the schema > ... RDF has a direct expression of type > ... elements ref'd in object properties can be embedded or > external > ... example asserts that record:AR123456 is a > fhir:AdverseReaction, as well as a SNOMED 241931004 > ... protege threw in that it was a named individual > ... do we talk about a symptom as an AdverseReaction? > ... in RDF, we don't care if a node is local or needs to be > resolved via a query > > dbooth: record: is instance data? > > Marc_Twagirumukiza: this SNOMED code is a big class. do we > expect to have granular codes e.g. skin rash? > > Tony: snomed equiv need not be expressed in the instance. > ... the relationshop between fhir:AdverseReaction and the equiv > snomed term would not appear in the instance data > ... [slide 7] > ... defining constraints about an "Adverse Reaction" > ... this is a style for doing a FHIR ontology > ... [slide 8] > ... this is a slice of a snapshot of the SNOMED ontology > ... there are versioned variants (seen at the bottom) > > Cecil: if you're looking at an individual FHIR resource, why > would you carry the subclass? > ... could get verbose > > Tony: payload just has the ref to 241931004 > ... the rest is in the SNOMED ontology > ... [slide 9] > ... in ICD-11, codes are classes, as with SNOMED-CT > > Guoqian: ICD comes from WHO. WHO has produced an RDF rendering > of ICD-11Beta > > <dbooth> Mistake on Tony's slide: ICD-11 is from WHO, not ISO. > > Tony: do you have an example of the formation of WHO URLs? > > Guoqian: will see if i can share > > Tony: probably similar to IHTSDO's approach > > <dbooth> ACTION: Guoqian to figure out whether he can share URI > conventions for ICD-11 [recorded in > [23]http://www.w3.org/2014/11/25-hcls-minutes.html#action07] > > Tony: [slide 10] > ... when you import other ontologies, you see the larger > picture > ... notation: '?' indicates a value set > ... [slide 11 - combined ontology] > ... alergy instance in the middle. > ... all of this binding occurs with closures in the RDF > ... [slide 12 - example with allergy to penicillin] > ... when you say "some penicillin", it's idenifying an abstract > ... those may be better in the terminology > ... but when it's a particular instance, better in the instance > > Cecil: if you're using owl:someOf, your saying that it's an > unknown instance. > ... if i wanted to describe this, i'd put it in a > SubstanceAdministration resource with a lot ID/batch, etc. > > JoshM: when we see a FHIR representation, we see an instance. > where did that "some" come from > ... ? > > Tony: came from me playing with it, to get it to connect to a > SNOMED penicillin > ... this is an alergy to something that may happen in the > future > > Cecil: it's implied that it happened at some date or at least > that you recorded it on some date. > > <dbooth> Tony: record:SomePenicillin means a dose of penicillin > -- not the OWL notion of "some". > > Tony: [slide 13] > ... view of allergy 1 > ... infers that it's an alergic disposition and moderage > ... causative agent is inferred > ... goal was to represent this in OWL. > ... i didn't see any real problems > ... working from the UML concept model > > <dbooth> David: Did you translate the FHIR data manually to > RDF? Tony: Yes. > > Tony: saying "this is an artifact of the ITS XML representation > of FHIR, could it be different with RDF?" > > Marc_Twagirumukiza: can we use the SNOMED predicates outside of > SNOMED? > > <dbooth> Marc: The use of SNOMED as predicates. Can we use them > out of SNOMED terminology, in FHIR? > > dbooth: in theory those SNOMED predicates tie in to the rest of > the SNOMED ontology > > Tony: this was a big discussion in the earlier work > ... when do you bring in other stuff and when does it overlap > FHIR structure > > dbooth: have to make sure you don't contradict SNOMED so if you > import more SNOMED later, we won't violate anything > > daniel: this has been explored in CIMI > ... consider that any ontology with the ambition of SNOMED-CT > will be less expressive than an information structure. > ... in an ontology of health care, you'd represent universals > ... having laterality in the information model and the term > model is good but the terminology model will be less > expressive. > > dbooth: this is a motivation for step 6 in the yosemite project > > <dbooth> The need for consistency across ontologies is one of > the reasons for Step 6 of the [24]http://yosemiteproject.org/ > roadmap. > > [24] http://yosemiteproject.org/ > > <dbooth> ADJOURNED > > <dbooth> IRC info: [25]https://www.w3.org/wiki/IRC > > [25] https://www.w3.org/wiki/IRC > > <dbooth> meeting notes conventions: > [26]http://dev.w3.org/2002/scribe/scribedoc.htm > > [26] http://dev.w3.org/2002/scribe/scribedoc.htm > > Summary of Action Items > > [NEW] ACTION: Eric to establish/make a wiki page for C-CDA RDF > representations work [recorded in > [27]http://www.w3.org/2014/11/18-hcls-minutes.html#action06] > [NEW] ACTION: ericP to set up tracker [recorded in > [28]http://www.w3.org/2014/11/25-hcls-minutes.html#action01] > [NEW] ACTION: Guoqian to figure out whether he can share URI > conventions for ICD-11 [recorded in > [29]http://www.w3.org/2014/11/25-hcls-minutes.html#action07] > > [27] http://www.w3.org/2014/11/18-hcls-minutes.html#action06 > > [PENDING] ACTION: Kerstin and Ingeborg to prepare a status and > future state ideas for PhUSE-FDA work [recorded in > [30]http://www.w3.org/2014/11/18-hcls-minutes.html#action05] > [PENDING] ACTION: Rob and all to decide on a wiki for Term Info > work [recorded in > [31]http://www.w3.org/2014/11/18-hcls-minutes.html#action09] > [PENDING] ACTION: Tony and Rob to report their plan on > High-level concept mapping to RDF work [recorded in > [32]http://www.w3.org/2014/11/18-hcls-minutes.html#action08] > [PENDING] ACTION: Tony to find out more details about how iCat > handles ICD-11 ont and report back [recorded in > [33]http://www.w3.org/2014/11/18-hcls-minutes.html#action01] > > [30] http://www.w3.org/2014/11/18-hcls-minutes.html#action05 > [31] http://www.w3.org/2014/11/18-hcls-minutes.html#action09 > [32] http://www.w3.org/2014/11/18-hcls-minutes.html#action08 > [33] http://www.w3.org/2014/11/18-hcls-minutes.html#action01 > > [End of minutes] > __________________________________________________________ > > > Minutes formatted by David Booth's [34]scribe.perl version > 1.140 ([35]CVS log) > $Date: 2014-11-25 18:05:13 $ > __________________________________________________________ > > [34] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm > [35] http://dev.w3.org/cvsweb/2002/scribe/ > > 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 [36]http://dev.w3.org/cvsweb/~checkout~/2002/ > scribe/ > > [36] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/ > > Guessing input format: RRSAgent_Text_Format (score 1.00) > > Succeeded: s/slive/slice/ > Succeeded: s/CID-11/ICD-11/ > Succeeded: s/some/"some"/ > Succeeded: s/___/Marc/ > Succeeded: s/Work Projects/Review of Work Projects/ > Succeeded: i/zakim, RobHausem is really rhausam/Postponed approving last > week's minutes, because no ITS co-chair is on the call to permit formal > approval. > FAILED: s/<JoshM> So should I be dialing into gotomeeting, or the W3C br > idge?// > Succeeded: s/<JoshM> The channel topic indicates gotomeeting// > Succeeded: s/<JoshM> Ah// > Succeeded: s/<JoshM> I would like to suggest that talky.io may not work > for this particular group-chat use case.// > Succeeded: s/<JoshM> Is there a meeting happening somewhere? I tried bri > dge, talky, and gotomeeting// > Succeeded: s/<daniel> will audio be on gotomeeting as well?// > Succeeded: s/<trackbot> Sorry, but no Tracker is associated with this ch > annel.//g > Succeeded: s|s/<JoshM> So should I be dialing into gotomeeting, or the W > 3C bridge?//|| > FAILED: s/So should I be dialing into gotomeeting, or the W3C bridge?// > Succeeded: s|s/So should I be dialing into gotomeeting, or the W3C bridg > e?//|| > FAILED: s/So should I be dialing into gotomeeting, or the W3C bridge// > Succeeded: s|s/So should I be dialing into gotomeeting, or the W3C bridg > e//|| > Succeeded: s/Tony and all to decide/Rob and all to decide/ > Succeeded: s/please ignore the next hand to go up// > Succeeded: s/embodded/embedded/ > Succeeded: s/ht eearlier/the earlier/ > Found ScribeNick: ericP > Inferring Scribes: ericP > Default Present: +1.978.794.aaaa, DBooth, +1.617.500.aabb, +1.801.949.aa > cc, +1.202.260.aadd, EricP, +1.410.624.aaee, Joshua_Phillips, [IPcaller] > , +1.801.507.aaff, Tony, patricia, +1.415.537.aagg, StanHuff, CecilLynch > , rhausam, mscottm, Guoqian, +33.1.73.71.aahh, +1.617.500.aaii > Present: Tony_Mallia David_Booth Josh_Mandel Rob_Hausam EricP_Joshua_Phi > llips Patricia_Grime(sp?) Stan_Huff Cecil_Lynch mscottm Guoqian Charlie_ > Mead Marc_Twagirumukiza Rafael_Richards_(IRC_only?) Daniel_Karlsson Cati > _Martinez_Costa_(IRC_only?) > Got date from IRC log name: 25 Nov 2014 > Guessing minutes URL: [37]http://www.w3.org/2014/11/25-hcls-minutes.html > People with action items: all eric ericp guoqian ingeborg kerstin rob to > ny > > [37] http://www.w3.org/2014/11/25-hcls-minutes.html > > > [End of [38]scribe.perl diagnostic output] > > [38] http://dev.w3.org/cvsweb/~checkout~/2002/scribe/scribedoc.htm > > > > ************************************************************ > *********************** > Manage subscriptions - http://www.HL7.org/listservice > View archives - http://lists.HL7.org/read/?forum=its > Unsubscribe - http://www.HL7.org/tools/unsubscribe.cfm?email=lloyd@ > lmckenzie.com&list=its > Terms of use - http://www.HL7.org/myhl7/managelistservs.cfm?ref=nav# > listrules >
Received on Wednesday, 26 November 2014 08:20:31 UTC