RE: Wait a sec...What about the HL7 RIM An Universal Exchange Language

I would love to see some of this in RDF triples.  Personally I've had 
experience with RIM and relational DBs but not much at all in healthcare 
with triples.
My real concern, is that people will go off and re model healthcare from 
scratch without taking the RIM into consideration.
If someone has a RIM based triple store or representation of clinical data 
I'd love to see it.
I'm from Kaiser where we have about 9 million members and tons of data on 
most of them. We have Epic which uses Cache database, and we have lots of 
relational databases (mainly Teradata and Oracle).  We don't have any 
triple stores.  I'd love to see an EHR like system based on triples.  I 
wonder how it would perform with so much data?  In the meantime, we, as 
well as the majority of big players, are looking into health information 
exchanges using CCD (based on HL7 RIM and CDA) structures.  Parsing them 
into RIM objects and persisting them in RIM databases.

If there is a non RIM based, brand new canonical model for healthcare, it 
will set the clock way back to where we were in the 1990s.



NOTICE TO RECIPIENT:  If you are not the intended recipient of this 
e-mail, you are prohibited from sharing, copying, or otherwise using or 
disclosing its contents.  If you have received this e-mail in error, 
please notify the sender immediately by reply e-mail and permanently 
delete this e-mail and any attachments without reading, forwarding or 
saving them.  Thank you.




Michael Miller <mmiller@systemsbiology.org> 
12/17/2010 09:23 AM

To
Peter Hendler/CA/KAIPERM@KAIPERM
cc
conor-dowling@caregraf.com, public-semweb-lifesci@w3.org, 
public-semweb-lifesci-request@w3.org
Subject
RE: Wait a sec...What about the HL7 RIM An Universal Exchange Language






hi peter,
 
"You don't gain anything by decomposing it and recomposing it into RDF."
 
the scenarios where i have seen this work is where the data itself isn't 
touched, the application makes the transformation to RDF in memory then 
applies semantic tools.  so if i want to see if there are patients 
suitable for a clinical trial, a sparql query might describe the criteria 
quite nicely.  then i could run over EHR records, translating to RDF 
triples in memory then applying the query.
 
depending on the backing store, i've also seen this done successfully by 
translating the query to sql then turning the query result into rdf 
triples to answer the SPARQL query.
 
cheers,
michael
 
From: Peter.Hendler@kp.org [mailto:Peter.Hendler@kp.org] 
Sent: Thursday, December 16, 2010 4:37 PM
To: mscottmarshall@gmail.com
Cc: conor-dowling@caregraf.com; mmiller@systemsbiology.org; 
public-semweb-lifesci@w3.org; public-semweb-lifesci-request@w3.org; 
twclark@nmr.mgh.harvard.edu
Subject: Re: Wait a sec...What about the HL7 RIM An Universal Exchange 
Language
 
Just want to be clear about when we use the word HL7.  In the USA when you 
just say HL7 it is assumed you mean Version 2 which is in very wide use 
and is not OO at all.  You almost have to say RIM based or V3 before 
people assume you mean the OO model (RIM). 

There is a specific XML serialization of the V3 RIM called the XML ITS. 
 It generates the XML elements, attributes and the order that they must be 
serialized in. When you are dealing with a CDA document, then in addition 
to these rules there are "templates" that are mostly just verbal human 
readable rules that you must follow for a given template.  Templates refer 
to each other, so the C32 document mentioned in the Meaningful Use rules 
of the HITECH part of the ARRA act is made from HL7 and IHE and HITSP 
templates all referring to each other.  These templates specify many 
things including which vocabularies must be used (SNOMED for example for 
coding the diagnosis). 

For a C32 templated CDA you have no wiggle room. The XML is defined.  You 
don't gain anything by decomposing it and recomposing it into RDF. 
There is nothing to say that you couldn't come up with different ITS's 
 For example you could come up with a JSON ITS and I suppose you could 
also come up with another RDF ITS. I'm just not convinced it would add 
anything to the understandability and I think it would add a heck of a lot 
of size to the representation. 

 Because the diagnosis and much of the clinical data contained in a CDA is 
in SNOMED, I do see the benefits of using subsumption on the SNOMED codes. 
 For example you could ask for all the patients that had any form a 
diabetes or heart disease and then use subsumption to get the lists of 
possible SNOMED codes that would be in the records of the target patients. 
 What I don't see is the advantage of using RDF instead of the standard 
RIM XML syntax. 

You could come up with an entirely new model for healthcare not based on 
the RIM and based on RDF, but much of the world (outside the USA) is 
already using the RIM. 






NOTICE TO RECIPIENT:  If you are not the intended recipient of this 
e-mail, you are prohibited from sharing, copying, or otherwise using or 
disclosing its contents.  If you have received this e-mail in error, 
please notify the sender immediately by reply e-mail and permanently 
delete this e-mail and any attachments without reading, forwarding or 
saving them.  Thank you.





"M. Scott Marshall" <mscottmarshall@gmail.com> 
12/16/2010 03:47 PM 


To
conor dowling <conor-dowling@caregraf.com> 
cc
Michael Miller <mmiller@systemsbiology.org>, Peter 
Hendler/CA/KAIPERM@KAIPERM, twclark@nmr.mgh.harvard.edu, 
public-semweb-lifesci@w3.org, public-semweb-lifesci-request@w3.org 
Subject
Re: Wait a sec...What about the HL7 RIM An Universal Exchange Language
 








I like Eric Neumann's description of RDF as "recombinant data".  

Agreed. Choosing something other than HL7 as the lingua franca for 
assertions doesn't devalue HL7! We can be happy that we got the 
information from one machine to another! It's a long haul from the days of 
big-endian, little-endian. The value is not in the messages (or message 
syntax) but what is in them (the cargo, the payload). But how will we 
interoperate between HL7 and CDISC? I suppose that an ontology will help.. 
BRIDG anyone ;) Cecil? 

The way that XML quietly infiltrated all our computer systems was by 
making it easy to describe and parse data of all shapes and sizes. Will 
OWL/RDF do the same by making it reasonably easy to describe the meaning 
of messages and documents? 

HL7 isn't going away. It is the standard. So, how can its users take 
advantage of other (non-HL7) sources of information that are related to 
the contents of its messages? And how can other systems, for example, 
clinical research systems relate their information and constraints to HL7 
data? See http://hcls.deri.org/coi/demo/ (makes use of pseudo-CDISC and 
HL7, and Drug Ontology), presented at AMIA. There should be 
machine-readable and reason-able links from one set of assertions to the 
other, that can make use of context (read: provenance). Could the HL7 
provenance help us make use of the 'cargo' in another context? i.e. 
assertion came from message issued by..  

-Scott 

-- 
M. Scott Marshall, W3C HCLS IG co-chair, http://www.w3.org/blog/hcls
Leiden University Medical Center / University of Amsterdam
http://staff.science.uva.nl/~marshall 

On Thu, Dec 16, 2010 at 11:06 PM, conor dowling <
conor-dowling@caregraf.com> wrote: 


On Wed, Dec 15, 2010 at 8:47 AM, Michael Miller <
mmiller@systemsbiology.org> wrote: 
hi all, 
  
"unambiguous identifier for "things"" 
  
i agree, this has been a known issue for many years (as you well know, 
tim) but its importance is just now growing as multi-omics studies and 
sharing of EHR records is becoming more common. 
  
"It is HL7 V3" 
  
i also agree, in a sense, with this.  HL7 messages capture information as 
a whole, as an entity, so in that representation it is also true that 
semantic web technologies would have a hard time, as is, making sense of 
them because semantic web technologies wants a fact by fact 
representation, e.g. triple store. 

But turn this on its head. HL7 messages come from "islands of data" which 
have undetermined linkage. Think of a lab result that has a local code, 
rather than LOINC. LOINC is equivalent to a link to the outside. 
Effectively the local code is meaningless outside. 

By its nature, linked data should resolve. If there is a url, you should 
be able to chase it down. The equivalent of a local code is a resolvable 
URL which presumably leads to some sort of description of what that local 
concept means, perhaps enough to translate it to a more commonly 
understood equivalent. 

You ask for any number of triples from a semantic endpoint, enough to 
capture what you need - all lab result assertions over a period for 
such-and-such a person. That's no different than a query in HL7 (or any 
other RPC like mechanism). 

The key difference with linked data (specifically) and "islands with 
protocol access" is linkage: the idea that links always resolve to 
something meaningful as opposed to identifiers that while unambiguous, may 
lead you no where. The problem with the old school which Parsa's "30 years 
of XML and HL7 experience" captures nicely is wrapped up in this. 

I've coded this stuff a good bit and everyone gets fixated on the syntax 
of messages/xml blocks. People are happy if a coded element is "correct", 
that it "conforms" as opposed to being useful or meaningful. And the 
problem lies not with them, but the mechanism. It's put the focus on 
"truck", not "cargo". 

Conor 
  
cheers, 
michael 
  
  
From: public-semweb-lifesci-request@w3.org [mailto:
public-semweb-lifesci-request@w3.org] On Behalf Of Peter.Hendler@kp.org
Sent: Wednesday, December 15, 2010 8:18 AM
To: markw@illuminae.com
Cc: public-semweb-lifesci@w3.org; public-semweb-lifesci-request@w3.org; 
twclark@nmr.mgh.harvard.edu 

Subject: Wait a sec...What about the HL7 RIM An Universal Exchange 
Language 
  
The PCAST did not take into consideration (maybe they don't even know) 
there is an universal exchange language for healthcare.  It is HL7 V3. 
 The CDA is merely one of virtually infinite structures that can be 
constructed from the RIM.  The meta information as well as the clinical 
data is unambiguously represented by RIM.  There is no reason to ignore 
the thousands of man years that went into designing the RIM.  The RIM 
Based Application Architecture  (RIMBAA) work group at HL7 has had many 
demonstrations of RIM based applications.  We don't need to re invent the 
wheel.  CDA is only one particular RIM structure designed for one 
particular use case.  Those of us who have been working at HL7 for years 
are blown away by the suggestion that there needs to be a different wheel 
invented.


NOTICE TO RECIPIENT:  If you are not the intended recipient of this 
e-mail, you are prohibited from sharing, copying, or otherwise using or 
disclosing its contents.  If you have received this e-mail in error, 
please notify the sender immediately by reply e-mail and permanently 
delete this e-mail and any attachments without reading, forwarding or 
saving them.  Thank you.

Mark <markw@illuminae.com> 
Sent by: public-semweb-lifesci-request@w3.org 
12/14/2010 06:44 PM 
 


To 
"Tim Clark" <twclark@nmr.mgh.harvard.edu> 
cc 
public-semweb-lifesci@w3.org 
Subject 
Re: An Universal Exchange Language
  
 








But seriously, Tim, if we were to pursue this problem, we would need some 
 
form of unambiguous identifier for "things"... and given the distributed  
nature of the biomedical domain, we'd want to make sure that there was  
some way of resolving that identifier to obtain metadata about it from a  
variety of disparate sources who might have very different information -  
clinical, molecular, demographic, etc...

hmmmm....