On 23/04/2007, at 12:30 PM, Tim Churches wrote:
David Guest <[EMAIL PROTECTED]> wrote:
David More wrote:
Hi David,
I don't - but would want to want the decision support / knowledge
management axis to the other capabilities.
DS is nice but nobody has shown you can do that yet. It might be
feasible with data codification but there are lots of implementation
issues before even that happens. I'd go for the low hanging fruit.
Agree. Decision support systems (DSS) should be the subject of much
funded research and field trials, but it is not ready for prime-
time because we don't have a good way of unambiguously representing
medical information in a computable (coded) form, except in small
niche sub-domains. Try building, say, a decision support system for
antibiotic prescribing. The clinical problem/diagnosis/setting
information is not encoded adequately in *any* existing clinical
information system, and microbiology results are currently reported
by labs as free text, expressed a zillion different ways. Sure you
can build a system in the research lab which the clinician re-keys
the relevant data into the DSS tool, but that will be used about
once in practice before users become tired of re-typing data that
is alreday in the system, but in the wrong form.
But gee, wouldn't an open-source primary care reference IT platform
would be a great way to turbocharge applied research and field
trials into the use of SNOMED CT and classifications/codesets in
real-life settings, and research and field trials of DSS built on
top of that? As it stands at the moment, any academic group wishing
to engage in such trials needs to negotiate on a commercial basis
with the major closed-source provider(s) of GP clinical info
systems, which is why we are seeing so little field work in this
area. Australia can get a jump on the rest of the world here, if it
moves swiftly.
Hi TimC,
I agree with your diagnosis that unambiguous computer representation
of complex health data for decision support is a potential show
stopper. Am not so sure that piling higher and deeper into a numeric
coding model will be a sustainable solution in the long run. I
digress here, the following story may interest you. I was plowing
through John Wood's "Leaving Microsoft to change the world" and his
adventures in founding the Room To Read organization. On page 182 he
talks of "But for the Nepali women, the situation is even worse, with
75 percent of adult women unable to read or write a simple
sentence." Back to the main thread. Could this quote hold the key to
the sorry state of decision support/e-health. The key is "A simple
sentence" , or rather the ability to string together a collection of
simple sentences that makes sense - is the hallmark of literacy.
Using a model (fixated on vocabulary) based on millions of concepts
assigned a unique number each, but without a syntax and grammar -
the EHR systems we have been building are functionally illiterate
like the women described by John Wood. What are the examples of such
simple sentences in healthcare that we need to code ?
This guy/gal has a family history of diabetes mellitus - diabm:fh
This guy/gal has diabetes mellitus as one of his list of problems
- diabm:eval
This guy/gal has chronic renal failure from diabetes mellitus -
crf:eval,from:diabm
This guy/gal has a right breast lump 2 cm on physical examination -
[EMAIL PROTECTED]:px,val:2cm,ctx;righ
This guy/gal has a cough for 2 weeks - coug:hx,for:2/52
This guy/gal has been losing weight for 6 months - [EMAIL PROTECTED]:hx,for:
6/12
This guy/gal is on gliclazide for diabetes mellitus -
glic:rx,for:diabm
This guy/gal is on gliclazide 30mg once a day - glic:rx,dose:30mg,qty:
1,freq:1/7
Docle closures** which are "context complete clinical codes" , are
the "simple sentences" referred to by John Wood.
As to moving swiftly and getting a jump on the rest of the
world....we have already moved...to docle closures generated from
natural text, working with Ruby's string handling.
Disclosure policy: Docle closures are increasingly used in the
medical director system.
HTH
Kuang
**A docle closure is a composition of base docle codes, that is a
complete representation of a medical proposition and good enough for
decision support.
_______________________________________________
Gpcg_talk mailing list
[email protected]
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk
