David Forslund [mailto:[EMAIL PROTECTED] wrote:I guess Dave was talking about raw page building, whereas I think Tim is assuming quite high level tools - like OIO or what we envisage with screen building based on archetypes and templates, which won't be that far off...but certainly the general principle of having clinical people drive their own systems, without the obstruction of the poor understanding of IT types I think is the only way to go. Horst mentioned in a post here somewhere the problem of IT people never delivering what was needed, and usually over time and over budget. This is the story of IT, almost globally, and for nearly its entire history. It stems from the mad idea (and it took me a few years to realise this - I'm not talking flippantly here) that somehow software people were going to be able to write the requirements specifications, designs, implementations, test plans and deploy systems that would _really_ do what the user group wanted - with at best, facilitated use case discussions and a few reviews. Most domain users don't understand UML, state diagrams etc, a few understand technical use cases...and why should they. What they do understand is: their own language, their forms, their workflows. Domain experts understand and develop their own ontologies.
I do believe that we shouldn't have physicians building web pages.
Huh? Surely the aim is to have all clinicians building the information displays, representations and data capture systems they require, using sophisticated tools which help them properly
structure that information (eg openEHR) and tools which eliminate
the need for arcane knowledge of programming languages. The new
batch of clinicians are beginning to embrace their role as knowledge
workers and informaticians. Not as programmers, sure, but the need
for programming is a transitional annoyance...
The history of IT in my opinion has taught us that the only kind of useful system that can be delivered to domain users (excepting where the domain itself is very technical/engineering oriented) is one which is not just configurable, but "programmable" - not by statements of source code, but by high-level _domain-user-oriented_ tools. I think we are making steps along this road. OIO is certainly this category of system; as our work continues in archetypes and templates, these things will be the basis of another generation of this category of system.
It may interest people to know by the way, that our work on ADL (Archetype Definition Language) is proceeding well, and we will spend some intense sessions at HL7 memphis discussing how to integrate its idea with HL7 templates, OWL and so on. We will have some kind of tool available in an early release soon, and we are hopeful of further support to develop this (all open source of course;-). ADL will also be presented as a main part of the proposal to do archetypes in the CEN ENV 13606 revision, in Denmark at the end of September. The archetypes/templates work could have been technically achieved more quickly, but we have judged that wide buy-in and standardisation are crucial to it being any more than a niche approach.
- thomas
