I have been waiting to see Tim's reply to Hugh's message and now wish to
add my own observations. While my comments show a different experience
in dealing with OpnEHR I do not wish to be seen as hostile to it. I
think it has much of merit but I do have some negative experiences.
Last summer we decided to get an understanding of OpenEHR and ran a
project to analyse their materials. Our conclusion was that the data
model was so complex that retreivals are likely to both perform slowly
and be so complex that maintenance would be very difficult. This report
was presented as a paper at HIC2006 - to my knowledge no-one from
OpenEHR interests attended the presentation apart from one person who
came up to me after the presentation and claimed our conjectures were
wrong and that he had used it for an implementation.
Subsequently, we ran another project to implement an OpenEHR system for
patient records. We used the materials available from the OpenEHR web
pages. We found that we couldn't actually complete the system as there
was necessary knowledge held by the staff that was not avaiable to us,
not even when we requested it. SO we have a 90% completed system. Our
objective had been to construct a system which we would then use to
construct a large synthetic data set so that we could implement a real
system and stress a persistent storage model. Obviously we never got to
create the synthetic data, build the persistent store or do the stress
testing.
I went to the OpenEHR stand at HIC2006 and engaged in disucssions with
their people. All but one was clearly nervous/hostile. The staff
protested that they did have a persistent store and I asked for it to be
demostrated. In those demonstrations it become apparent that by
persistent store they meant storage of the archetype data model - not
patient records - they said that wasn't their concern.
These experiences have left me with a dubious image of OpenEHR progress.
So I am left very much in the same position as Tim and basically concur
with his commentary. Particularly, if Hugh's protest below is to be
taken as legitmate then test systems need to be available for stress
testing, and we would be quick to put up our hands to do the testing.
Meanwhile we must get on with our work and are researching our own
architecture for hospital information systems based on a number of
criteria, the most important of which is to give the greatest weight to
the users mental models of processes and workflow so that they can be
mimiced as closely as possible in the eletronic system. We are writing
up this work and hope to release it shortly. The prototype demonstrates
the essentials of data capture and automatic workflow of data movement
from one station to the next.
cheers
jon
Hugh Leslie wrote:
Hi Tim
I wish you would sit down with us and have a look at what we are doing
rather than 'be a cynical bastard'! We have appreciated your support over
the years and I can understand your frustration, however we have been
working really hard with few resources - most of the software described
below is the product of small teams working with very small budgets. Its
only really in the last 12 months since version one was released that all
this has happened.
I have included part of the text of the email I sent to you recently and
again would be happy to sit down with you and show you where we are at. ALL
of this software was demonstrated at the Ocean stand at HIC this year,
including taking HL7v2 pathology and radiology results and storing them as
openEHR compositions in the database.
We as a company at the moment are more concerned about having too much work
and too much interest rather than not enough. Most of this is in Europe,
and I for one am proud to be part of an Australian company that is doing
world recognised work in this area.
"We currently have demonstrable software that:
- creates archetypes (the ocean archetype editor and a java archetype
editor).
- We can take the archetypes and create templates from them which you can
use to build gui forms using a drag and drop process. (Ocean Template
Designer)
- We have worked with clients who have taken these guis and used them to
enter data and save the data into our repository. (Ocean EhrBank)
- EhrBank is an openEHR repository which will store any archetyped data
(the ocean kernel sits on top of it providing the ability to build openEHR
objects from archetypes or templates.
- We have a query editor that allows you to build sql like queries to query
the data and we have clients in Turkey that have successfully demonstrated
this functionality for themselves.
- We have software that allows you to display data queried from the
repository in a web page.
We also have a terminology service that is well integrated into the tools
and allows you to build pre-defined queries in snomed etc. These can then
be used in a gui to allow clinicians to select terms.
We are not the sort of company that hypes up its wares and so Tom is being
honest with you in saying that these systems have not been tested in high
volume mission critical situations (although volume testing of the
repository is going on as we speak.) These tools have been out there being
demonstrated and tested by real clients for at least 6 months and we have
generated a lot of interest across Europe (In particular Turkey, the
Netherlands and Denmark and there is a lot of interest in the UK).
Most of these tools have been built since openEHR v1.0 was released so in
fact the timelines have been particularly spectacular in my opinion
especially since we have done most of it under our own steam. We can
demonstrate the full chain that you referred to but not all of it is as
slick as we would like it yet (that said, we think its all pretty slick!)
If you were at HIC this year, you would have seen this all working at the
stand, and we have come a long way since then.
We are definitely ready to start some small implementations and there are
opportunities in Australia as well as Europe (and the US as Sam mentioned.)"
Apart from these things, Tom and I were invited to an eclipse meeting in
Europe recently to discuss an eclipse version of the openEHR tools and this
process is on-going.
Regards Hugh
I forgot to mention that there is the beginnings of an archetype repository
available from the openEHR web site and the Ocean web site.
M: 0404 033 767 E: [EMAIL PROTECTED]
-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Tim Churches
Sent: Thursday, 14 December 2006 8:50 AM
To: General Practice Computing Group Talk
Subject: Re: [GPCG_TALK] Request for XML dump for RAIL
David Guest wrote:
The data is identified as such and grouped in a hierarchy. But its
description and constraints are entirely open and
undefined. I guess
its a useful lowest common denominator
Does anyone think openEHR will ever produce the goods?
There has been correspondence on the openhealth mailing list
regarding this issue recently. David More quipped about
"geological timescales", perhaps with some justification. To
summarise and paraphrase (accuratey I hope) the thread: the
openEHR people assure us that several private firms are using
openEHR-based systems in deployed proprietary vertical health
apps, and that lots of profs and students in various
universities are studying and tinkering with it. The openEHR
specifications have been accepted as a proposed standard (but
not ratified or approved as a standard as yet). Furthermore,
Ocean Informatics and the openEHR Foundation are themselves
working on a suite of tools which actually implement the
ideas behind openEHR, but these tools are in different stages
of completeness: tools to define and edit openEHR archetype
definitions are complete and available as open source. Tools
to actually store and retrieve data using openEHR archetypes
are at alpha or beta stages in the openEHR secret laboratory,
but have not been fully tested and are not ready for
production use. Thomas Beale has offered access to an openEHR
engine hosted in the Ocean Informatics labs, to be accessed
via a proprietary Web service interface requiring the use of
a Microsoft C# .NET DLL on the client side, for
capability-testing purposes by interested parties (contact
Thomas Beale at Ocean Informatics if you are interested). All
these openEHR tools still under development may or may not be
open sourced in the future - the Ocean Informatics and
openEHR people need to investigate business models. Other
parts of the openEHR puzzle, such as a shared library of
openEHR archetype definitions, and a full query language, are
still on the drawing board or in only early stages of
implementation. Oh, there is also an open source version of
an openEHR storage/retrieval kernel being written in Sweden,
but it is not yet complete either.
I asked the same questions of the Ocean Informatics and
openEHR people in 2003, and after much email correspondence
and head scratching, I was assured that usable,
production-quality openEHR implementations would be available
quite soon. The same assurances were given just a few weeks
ago. I conclude that they are indeed a bit further along now
with actual implementation than they were three years ago,
but still have quite a way to go, but it is very hard to
extrapolate the progress line to divine when it might cross
the V1.0 boundary, although the fact that they were working
on GEHR (the predecessor to openEHR) about 15 years ago, and
the openEHR has been going for nearly a decade might provide
some clues. Perhaps the remaining distance is being halved
with every passing year? Or perhaps I am just a cynical bastard?
Tim C
_______________________________________________
Gpcg_talk mailing list
[email protected]
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk
__________ NOD32 1920 (20061213) Information __________
This message was checked by NOD32 antivirus system.
http://www.eset.com
_______________________________________________
Gpcg_talk mailing list
[email protected]
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk
begin:vcard
fn:Jon Patrick
n:Patrick;Jon
org:Faculty of Engineering;School of Information Technologies
adr;dom:;;University of Sydney
title:Chair of Language Technology
x-mozilla-html:FALSE
url:http://www.it.usyd.edu.au/~jonpat
version:2.1
end:vcard
_______________________________________________
Gpcg_talk mailing list
[email protected]
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk
