At 1:36 pm +1000 21/7/06, David de Bhál wrote:
Is there any documentation of this "free utterance" idea.
Surely language and syntax of the user is a major factor in software
interpretation of the 'facts'.
Common knowledge in coding circles.
Documented in messaging standards.
Used by the GP Vocabularly project (Don Walker,
UAdelaide) to develop the prototype GP term set.
So I'm sure someone could find papers if needed.....
Peter Elkin a couple of years ago at na HL7
meeting in Melbourne reported research on coding
clinical documents completely in different coding
systems. Translation between different coding
systems was unreliable. In the final analysis,
text is bedrock. A coding system is a
representation of that.
Humans can understand it. Don't forget the human
still intermediates in the coding step (with some
machine assistance) - routine machine processing
of data stored that way uses only the coded
abstraction.
Ian.
David de Bhál
www.v-practice.com
________________________________
-----Original Message-----
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]
On Behalf Of kuang oon
Sent: Friday, July 21, 2006 10:32 AM
To: General Practice Computing Group Talk
Subject: Re: [GPCG_TALK] SNOMED Project Proposal
G'day,
I agree with Ian's comments. "Free text utterance" means more
expressivity and simultaneously the need to capture multifaceted
clinical information. We can embede / link numerical values such as
clinical duration, frequency and magnitudes of all sorts into
codes . Using a text editor for input, rather than from a pick list
- is desirable and doable.
Free text entry can solve the problem of coding for useful clinical
scenarios that stretch across a plurality of codes that explodes in
a combinatorial fashion. Useful examples of such combinations are i)
an illness - a drug treatment - patient's outcome to the drug
treatment pattern and ii) test - test result pattern. A code for
each conceivable scenario is paving a road to enumerated
combinatorial hell.
BTW I have been looking at Duncan's 'troponitis' - it is really 3
atomic concepts: 1)the serum troponin test 2)the test result and 3)
negative criteria - in short a classic molecular concept code. One
can just assign an arbitrary number code for "troponitis". Here
we have the fork in the road. In an emergent coding system, such
as DOCLE, you have to make more from less.
Using this refactored solution, the troponitis code is constructed
pro re nata from a palette of already available docle codes.
Hence by feeding a tiny fragment of natural language oriented SHEEP
thru its parser :
ModelSheepService new sheepParse:
'[sh
problems
troponin find: abnormal high no: acute myocardiac infarction
sh]'
gives doclescript output of:
OrderedCollection ('&[EMAIL PROTECTED]@trop-onin],find[abnormal,high],no
[myocardialInfarction]')
with the canonizer gives the docle code for troponitis:
[EMAIL PROTECTED]:eval,find:abno,high,no:myoci
From the programming viewpoint, the docle code is amenable to 1)
pattern searches on SQL 2)be exploded into their respective atomic
concepts 3)visual confirmation of veracity of the code generated
from natural text. The docle code is auto-reflective and explains
its own meaning.
Note also that "free text entry problem " and "level 4 ehr
interoperability problem" are the same or essentially the same as
they both can be addressed from a single SHEEP perspective.
HTH
kuang
P.S. BTW Duncan, your cardiology input is appreciated and never
taken for granted. The use case for "troponitis" is as real and
relevant as it gets.
On 20/07/2006, at 10:48 PM, Ian Cheong wrote:
Actually, the preferred method is to record the "free text
utterance" of the clinician and attach a coded term to that.
Recording only a coded term loses richness and therefore usefulness
> for human decision makers.
Text can be coded after the fact with well designed tools.
Loss of specific information is much harder to recreate.
Ian.
At 5:20 pm +1000 20/7/06, Greg Twyford wrote:
Michael Tooth wrote:
Peter Machell wrote:
On 20/07/2006, at 12:13 PM, Greg Twyford wrote:
GPs are already coding data in programs like MD. It's
diagnosis/ past
history dialogues have forced them to consciously avoid using the
available coded entries if they don't want to use them, and enter
uncoded stuff instead.
Most of the MD users I see enter free text for all notes, and
are offended when I suggest they first use the 'Reason' button.
Isn't it time that coded diagnosis be made the default entry
method, with free text a last resort?
You might remember that MD originally only had the coded option, but
that the ability to put uncoded in was really the result of there
not
being enough DOCLE codes.
Michael,
Originally it MD it was much easier for GPs to end up with their
own search entries like 'dia' in their notes, which were
meaningless in terms of coding, when trying to bring up the coded
diagnosis 'diabetes', for example.
The dialogue was changed to reduce the likelihood of this by
placing the free text box down the bottom and by changing the
action of the search entry box at the top. It's new action ensured
that one of the coded diagnoses was entered from the list when a
search entry like 'dia' was used in the default search entry box
at the top.
This effectively forced selection of a coded diagnosis/reason/past
history item unless the bottom 'free text' box was consciously
clicked on.
They also included the diagnosis coding tool in maintenance to
allow uncoded or meaningless entries in the database to be
corrected or linked with coded ones.
Greg
--
Greg Twyford
Information Management & Technology Program Officer
Canterbury Division of General Practice
E-mail: [EMAIL PROTECTED]
Ph.: 02 9787 9033
Fax: 02 9787 9200
PRIVATE & CONFIDENTIAL
*********************************************************************
**
The information contained in this e-mail and their attached files,
including replies and forwarded copies, are confidential and intended
solely for the addressee(s) and may be legally privileged or
prohibited
from disclosure and unauthorised use. If you are not the intended
recipient, any form of reproduction, dissemination, copying,
disclosure,
modification, distribution and/or publication or any action taken or
omitted to be taken in reliance upon this message or its
attachments is
prohibited.
All liability for viruses is excluded to the fullest extent
permitted by
law.
*********************************************************************
**
_______________________________________________
Gpcg_talk mailing list
[email protected]
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk
--
Dr Ian R Cheong, BMedSc, FRACGP, GradDipCompSc, MBA(Exec)
Health Informatics Consultant, Brisbane, Australia
Internet: [EMAIL PROTECTED]
(for urgent matters, please send a copy to my practice email as
well: [EMAIL PROTECTED])
PRIVACY NOTE
I am happy for others to forward on email sent by me to public
email lists.
Please ask my permission first if you wish to forward private email
to other parties.
_______________________________________________
Gpcg_talk mailing list
[email protected]
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk
_______________________________________________
Gpcg_talk mailing list
[email protected]
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk
--
No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.1.394 / Virus Database: 268.10.3/394 - Release Date: 7/20/2006
--
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.1.394 / Virus Database: 268.10.3/394 - Release Date: 7/20/2006
_______________________________________________
Gpcg_talk mailing list
[email protected]
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk
--
Dr Ian R Cheong, BMedSc, FRACGP, GradDipCompSc, MBA(Exec)
Health Informatics Consultant, Brisbane, Australia
Internet: [EMAIL PROTECTED]
(for urgent matters, please send a copy to my
practice email as well:
[EMAIL PROTECTED])
PRIVACY NOTE
I am happy for others to forward on email sent by me to public email lists.
Please ask my permission first if you wish to
forward private email to other parties.
_______________________________________________
Gpcg_talk mailing list
[email protected]
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk
