W3C home > Mailing lists > Public > public-semweb-lifesci@w3.org > January 2013

Re: HL7 RIM Designtime OWL Runtime RDF

From: <Peter.Hendler@kp.org>
Date: Wed, 16 Jan 2013 14:41:33 -0800
To: mccusj@rpi.edu
Cc: conor-dowling@caregraf.com, david@dbooth.org, d.rebholz.schuhmann@gmail.com, jluciano@gmail.com, meadch@mail.nih.gov, michel.dumontier@gmail.com, public-semweb-lifesci@w3.org, ri@semanticidentity.com, rmrich5@gmail.com, sivaram.arabandi@gmail.com, tfmorris@gmail.com
Message-ID: <OF93597906.95564C10-ON88257AF5.007AE9B7-88257AF5.007CA6C7@kp.org>
This is a special use case.  Although there is HL7, there are also other 
standards like openEHR, CIMI, and there are proprietary models for 
clinical information.  We can be talking about a document, a message or a 
"clinical statement" which is a fragment of one of the above.
Even within one standard (like HL7 v3), there is no agreement on what part 
of the model "always" get's the negation.  You are allowed to negate a 
class like an Observation, or you can instead, not negate the Observation 
but use a SNOMED code, pre or post coordinated that is negated.  The 
bottom line is you may have different syntactic legal HL7 or other 
"standard clinical statement models" that "mean the same thing".  A human 
can read the models and make a judgement if they mean the same thing 
(semantically equal) but we always talk about a machine way to test for 
this.

It hasn't been solved or done.  Part of the problem is that the different 
parts of the clinical model aren't even based on the same open or closed 
world assumption. The Observation Act (OO Class) is closed but the SNOMED 
diagnosis code is part of the SNOMED EL+ logic open world.  The negation 
can be on either part and it could mean the same thing.

A negated Observation of Pneumonia may be the same thing as an Observation 
of NOT pneumonia.

Is ther a way to normalize these different representations to see if they 
are the same?  I think maybe.  Maybe if you are allowed to "freeze" them, 
declare them final and closed, and then decompose them into RDF triples. 
Then you can see if the two graphs are the same.  I haven't done it. I 
haven't tried it, I don't know anyone who has.  I'm wondering if there are 
"obvious" reasons that it won't work that I don't know.

Once the different clinical statements have been sent, they are frozen. 
You are not allowed to change them. So I'm guessing and hoping that from 
that point on, you can decompose them into a closed set of triples and 
then see if they overlap or are the same.

A machine way to tell if two syntactically different clinical models, both 
composed of HL7 RIM and SNOMED are in fact representing the same thing.

Maybe it depends on your interpreatation.


1  This person     Is observed to have           a diagnosis of Pneumonia

Now two kinds of negation

2  This person    Is NOT observeed to have         a diagnosis of 
Pneumonia
3 This person    is observed to have                   a diagnsosis of NOT 
pneumonia

Maybe the triples don't really help.  This might be a judgement call of 
whether these two things are "for clinical purposes" close enough to be 
considered the same thing.

To a clinician these both might just means  "This person doesn't have 
pneumonia".











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.




picture
(image/jpeg attachment: 01-part)

Received on Wednesday, 16 January 2013 22:42:20 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 18:01:17 GMT