Dear Andrew and others,
I must take responsibility for the metaphor of gold, silver, bronze (platinum is Gerard's addition!). My intention was not to indicate a first, second, third quality of achievement as in the olympics, but really to indicate level of completeness/sophistication. I did not wish to imply that simple models are inferior, they just serve different purposes from complex ones. I would like to have found an alternative metaphor but could not think of one when I was writing to Peter several days ago. I would still be grateful for suggestions.
Andrew, you do raise an important point about what can be cut out. I will copy here a paragraph I have just sent around the Reference Model Task of the CEN Task Force on this.
"Through openEHR David Lloyd and I have been working to build on a decade of practical EHR experience from GEHR, EHCR-SupA, Synapses, SynEx and 6WINIT resulting in a working ENV13606-like record server using a dual model approach, combining this with a similarly rich experience in Australia and catering for new requirements in areas such as text and language processing, decision support and bioinformatics, to arrive at a “best of breed” approach to the Reference Model and Archetype Model, ready for a new wave of implementation and validation. We recognise that the openEHR models now therefore contain some features that need new evaluation, but many features are fairly well tested approaches that could be adopted for standardisation."
We must accept that, whilst we believe the openEHR models to be the best basis on which to build the next generation of EHR servers, we do not have implementation experience to back all parts of it. It is likely that some refinements will emerge as necessary in the light of implementation deployment and live clinical use. I therefore suggested that, whilst openEHR has a legitimate desire and drive to evaluate next-generation solutions, a legislative standard ought to build on those aspects which are most solid and can be regarded as reasonably stable. openEHR needs to be free rapidly to evolve on the basis of new requirements and implementation experience. The openEHR approach and models exist in their own right, and can be taken up by any vendor or demonstrator without endorsement from CEN.
The new CEN standard (for EHR communication) does need to be compatible with openEHR, and it should be possible for a standards-conformant vendor easily to migrate from new-CEN to full openEHR, but I don't think we will do CEN or openEHR a service by seeking to have these two exactly tied together at any one point in time.
However, if you accept that premise (and you are welcome to disagree) the question remains about how to derive a suitable candidate for standardisation from openEHR. David and I have begun working in this area, with offer of help from Tom. I think we would welcome ideas from the broader openEHR community about the methodology we should pursue.
With best wishes,
Dipak
