W3C home > Mailing lists > Public > public-linked-json@w3.org > February 2015

Re: JSON-LD for emerging healthcare standard (FHIR)?

From: David Booth <david@dbooth.org>
Date: Tue, 24 Feb 2015 16:52:13 -0500
Message-ID: <54ECF28D.1000709@dbooth.org>
To: Josh Mandel <Joshua.Mandel@childrens.harvard.edu>, Nicholas Bollweg <nick.bollweg@gmail.com>
CC: "public-linked-json@w3.org" <public-linked-json@w3.org>, Grahame Grieve <grahameg@gmail.com>
Hi Josh,

Yes, these were both factors that Lloyd McKenzie and I discussed on 
today's teleconference:
http://www.w3.org/2015/02/24-hcls-minutes.html#item03
A primary consideration is to figure out the absolute *minimal* changes 
needed to turn FHIR JSON into workable JSON-LD.  Ideally it would 
involve only the addition of a single, standard "@context" element, but 
as you noted (and I noted on today's teleconference), it looks like some 
additional tweaking would be needed beyond that.  The goal is to figure 
out exactly what those tweaks would have to be, so that the change can 
be adequately considered.  If this plan does not pan out -- or if we run 
out of time -- the fallback would be to define a separate RDF 
representation for FHIR, which might be serialized in Turtle, JSON-LD or 
any other RDF serialization.

Thanks,
David

On 02/24/2015 03:19 PM, Josh Mandel wrote:
> Yes, Nicholas said this correctly: SMART did use JSON-LD for our "SMART
> Classic" API, but we've moved away from this with the transition to
> FHIR. For what it's worth, two points of caution with respect to David's
> plan:
>
> 1. I think if you dig into the FHIR JSON format, you'll find that since
> it takes its approach to references, and re-ification of some URIs as
> "Codings" (with "system", "version", and "code" properties), etc. it's
> not the easiest thing to try to translate into idiomatic JSON-LD. (That
> is, you probably can't just layer on a context and be done with it.)
>
> 2. Just as a reality check: current trajectory of FHIR's standardization
> process is not in a place where it's likely to accept a broad-scale
> replacement of the serialization format between now and the publication
> of the 2nd draft (DSTU2).
>
> I don't want to discourage experimentation here. Just want to set
> expectations appropriately, especially around the timing. If we really
> want to think about making this work, the best plan would probably be to
> consider a distinct JSON-LD serialization for FHIR, rather than trying
> up-front to replace its existing JSON format.
>
> Best,
>
>    Josh
>
> On Tue, Feb 24, 2015 at 11:44 AM, Nicholas Bollweg
> <nick.bollweg@gmail.com <mailto:nick.bollweg@gmail.com>> wrote:
>
>     Had investigated this some in the past.
>
>     There's this...
>     http://docs-v06.smartplatforms.org/framework/json-ld.html
>
>     Though it makes it sound like they are moving off JSON-LD to FHIR else?
>
>     Nevertheless, I am certain there would be interest in pursuing this.
>
>     On Tue, Feb 24, 2015 at 2:29 PM, David Booth <david@dbooth.org
>     <mailto:david@dbooth.org>> wrote:
>
>         Fast Healthcare Interoperability Resources (FHIR) is an emerging
>         healthcare standard being developed at HL7 that is gaining lots
>         of momentum, and currently has both XML and JSON
>         representations.  A joint HL7-W3C work group[1] has been formed
>         to promote the use of RDF for healthcare information
>         interoperability, and we've been investigating the potential use
>         of JSON-LD for FHIR instead of plain JSON, so that the same FHIR
>         JSON-LD messages could be automatically interpretable both as
>         plain JSON and as RDF (because JSON-LD is both JSON and an RDF
>         serialization).
>
>         This is potentially a fantastic opportunity to get JSON-LD
>         adopted in healthcare IT.  *However*, if the draft standards for
>         FHIR JSON are to be changed to specify JSON-LD, the proposed
>         changes need to be made ASAP -- in the next 2-3 weeks -- before
>         the FHIR standardization effort gets too far along.
>
>         I have started investigating the technical impact of switching
>         to JSON-LD -- see my interim report[2] -- but I'm running into
>         some issues around scoping and identifiers that I'm not sure of
>         the best way to solve.  I'm not a JSON-LD expert, so I'm
>         concerned that in my ignorance I may miss an important potential
>         solution.
>
>         Would anyone who is JSON-LD savvy be willing to help on this?
>         You do NOT need to know FHIR or anything about healthcare IT,
>         but your JSON-LD insight would be very helpful.
>
>         Thanks!
>         David Booth
>
>         1.
>         http://wiki.hl7.org/index.php?__title=RDF_for_Semantic___Interoperability
>         <https://urldefense.proofpoint.com/v2/url?u=http-3A__wiki.hl7.org_index.php-3Ftitle-3DRDF-5Ffor-5FSemantic-5FInteroperability&d=BQMFaQ&c=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU&r=c7b1QeR755-dBx2b0xnlepDTylromoEzcLl-6ixmBL3TpXSxSvtAvT553fzSgLpm&m=mktHFUWCTrWcrt1qk-iCjm47X5XfEVMpI_aBtD1fk3Y&s=MF0q3F2wjL-Bemcc66YRDZjXizyZcXQUBFs5fnfLhu4&e=>
>         2. http://dbooth.org/2015/fhir/__json-ld/fhir-in-json-ld.pdf
>         <https://urldefense.proofpoint.com/v2/url?u=http-3A__dbooth.org_2015_fhir_json-2Dld_fhir-2Din-2Djson-2Dld.pdf&d=BQMFaQ&c=qS4goWBT7poplM69zy_3xhKwEW14JZMSdioCoppxeFU&r=c7b1QeR755-dBx2b0xnlepDTylromoEzcLl-6ixmBL3TpXSxSvtAvT553fzSgLpm&m=mktHFUWCTrWcrt1qk-iCjm47X5XfEVMpI_aBtD1fk3Y&s=pkUAX-JkLZPOo349HKahcCPyvV18OEq0zJ82thPI-hI&e=>
>
>
>
>
Received on Tuesday, 24 February 2015 21:52:43 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 16:18:43 UTC