> Re: Fw: Interoperability with HL7 > On 10/02/2010 12:00, Andrew McIntyre wrote: > > I think a DCM format should exclude the administrative attributes, > such as Author and Observation Time > > Andrew, > > I could agree in principle, but how could Observation time be an 'adiministrative' attribute? > > The wording is probably wrong, its an attribute that is expected to be in the > information model and how its there is not something a DCM needs to worry about. >
well actually, I would agree with this, but whether such an attribute is in the information model depends on teh information model. This attribute happens to be in openEHR, along with other related attributes to do with Observation timing. This was precisely my point about why the reference model matters - what does it supply, what does it not supply? HL7v3 and 1s013606 don't supply Event.offset, or width for example.... so clearly DCM models have to assume a reference model, else they have no formal foundation. > A "Follow up date" might belong in the DCM. In reality its a line that has to be drawn based on > what is expected of an information model of the target systems. That is a vague line for sure but > if u look at the dominant few information systems it can be drawn easily for many attributes. indeed. This is exatly what DCM needs to do - to decide what 'information model' it is usiing. openEHR archetypes for example assume a reduced version of the openEHR RM. Any archetype has to assume someting, else it is all on shifting sands. > and leave those to the Information > model. Drawing that line is potentially tricky, but needs to be done > in order to allow each Information model to do it the way it wants to. well - I could agree with true administrative attributes, such as version & audit innformation. But attributes relating to observation timing, structure, order timing, conditionality, action timing & state machine states..... these are not administrative attributes. > Things like order numbers and links to orders should also stay out of > the DCM model. this might be a reasonable position, but on the other hand you might want to build an archetype that says that an order number is mandatory - how would you do this? > > In the end we want the pure clinical hierarchical structure that does > not conflict with the information model and ideally does not conflict > with the Terminology Model. it depends on what you mean by 'pure hierarchy' - if you literally mean node/arc, then you are forcing modellers to recreate basic data structures all the time, like 'list' used in Apgar, or the timing and state information used to define OGTT. These things are basic; if you force every DCM model to reinvent this, you are inviting a whole raft of errors and problems that don't need to exist. (more comments later - on horrible webmail right now) -- Thomas Beale Chief Technology Officer Ocean Informatics
