W3C home > Mailing lists > Public > public-semweb-lifesci@w3.org > March 2010

Re: semantic web for EHRs

From: <dirk.colaert@agfa.com>
Date: Fri, 12 Mar 2010 15:09:58 +0100
To: kidehen@openlinksw.com
Cc: conor dowling <conor-dowling@caregraf.com>, public-semweb-lifesci@w3.org, public-semweb-lifesci-request@w3.org
Message-ID: <OF0B73C7AB.6E73B870-ONC12576E4.004D115B-C12576E4.004DD139@agfa.com>
We (Agfa HealthCare) are doing a project on European level around 
infectious diseases. We collect data in different hospitals (with 
different CIS systems), we data mine the retrieved results sets , store 
the results in a knowledge repository and use this knowledge for decision 
The data collection, mentioned above is using sparql and a domain ontology 
we created. Having the results of the different hospitals in the same 
formalism and using the same ontologies ensure semantic interoperability 
and makes the merging of the results sets very easy.
The whole thing creates what we call a " virtual clinical data repository" 
because we don't have to centrally store clinical data. Each clinical site 
implemented a sparql endpoint, hiding the intrinsics of the specific site
We use D2R to convert from sparql to sql..

The site of our project is www.debugit.eu
Unfortunately there is no much technical detail on that site.

Kind Regards,

Dirk Colaert | Agfa HealthCare
Advanced Clinical Application Manager | HE/Advanced Clinical Applications 
T  +32 3444 8408 | F  +32 3 444 8401 | M  +32 497 47 08 71

Quadrat NV, Kortrijksesteenweg 157, 9830 Sint-Martens-Latem, Belgium
Click on link to read important disclaimer: 

Kingsley Idehen <kidehen@openlinksw.com> 
Sent by: public-semweb-lifesci-request@w3.org
12/03/2010 14:40

conor dowling <conor-dowling@caregraf.com>
Re: semantic web for EHRs

conor dowling wrote:
>     > U.S.? (There's little here from what I can see - the
>     interoperability push is around SOAP).
>     In my view, SOAP is the wrong direction.  It is just adds
>     complexity and
>     contributes to "babelization":
>     http://www.w3.org/2003/Talks/0717-semweb-dbooth/slide10-0.html
> you're right but here's the rub - there's $'s in babel. Bad IT - 
> translation layers and their maintenance - is good business, sometimes.
> Take the U.S. NHIN CONNECT project whose laudable goal is to allow 
> patient record exchange between institutions big and small. It tackles 
> what's need - security, credentials, opt-in etc - and then ... well, 
> it gets all SOAP'y. Gateways, adapters, layers, all those layers. What 
> about a "web of interlinked data", just add security policy ...??
> It's annoying because think how easy linking is - in reality and now, 
> not just conceptually, some time away. (I know I'm preaching to the 
> choir here but ...)
> Take a patient vital - http://vista.caregraf.org/rambler/120.5/716 
> (Christopher's blood pressure at a date). This record is typed by the 
> VA vital type, http://vista.caregraf.org/rambler/120.51/1 (blood 
> pressure), one of 19 that the system records ( 
> http://vista.caregraf.org/rambler/120.51 ). Vital type is a "locked 
> file" ( http://vista.caregraf.org/rambler/schema/120.51 ), one of many 
> terminology files in VistA.
> Now, on the face of it, such data is meaningless outside this VistA. 
> We need a "mapping layer", an "RPC". A "type-mapper". A reformatter. 
> Layers ...
> BUT WE KNOW (on this group) that it is trival to do something like ...
>                         :120.51/1 ---- same as -----> SNOMED:392570002
> and heh presto, your vitals are "linked". Were Christopher lucky 
> enough to end up in the Cleveland Clinic then this and his other data 
> would be trivial to query - no longer site or even VA-specific.
> And this isn't an isolated case. It's true in general. (I'm working on 
> an "linked patient browser" - needs very little code - and this 
> principle holds true for procedures, medicines, vaccines ...). 
> The train is leaving the station on health records (in the U.S. 
> 'meaningful use' is about to get nailed down) and they're made for the 
> web of data but all we have are soap bubbles, all a drift ...
Is there going to be an RDF model based Linked Data View over this data? 
Or are you looking for help re. Linked Data publishing etc?



Kingsley Idehen 
President & CEO 
OpenLink Software 
Web: http://www.openlinksw.com
Weblog: http://www.openlinksw.com/blog/~kidehen
Twitter/Identi.ca: kidehen 
Received on Friday, 12 March 2010 14:10:39 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 14:52:42 UTC