I’ve encountered another situation which I find hard to model based on archetypes and the RM.
we are building software for an ICU, and there is a form which is used to control significant physiological values over time.
Now if we think of one column of the table we will have a set of observations to be recorded at a precise point in time, which could be modelled as a Composition.
The problem with this situation is that the set of observations varies from time to time for the same patient and varies significantly from patient to patient in depending on the characteristics of the case (cardiology, nerological, trauma, etc.)
It seems quite clear to me that using archetypes which are predefined before runtime to model this compositions makes no sense here.
Is there any work around or is this kind of problem addressed with templates?
I imagine that this is the way you are using the archetypes. These should be used to model discreet and whole concepts - not data entry screens. The latter is made of a set of archetypes (which can be called at runtime). Templates are a way of gathering archetypes for use at runtime - we have a rudimentary form of templates running in our software - very simple and it works. Just says use this archetype and this one under this section - set the default value of this etc.
I also wonder if you are using the history model in observations - which arose from ICU requirements…
For each measurement you can have as many readings (events) as you wish all in one entry - or as many entries as you wish (where you want to record different sources of data) in as many compositions as you wish (medico-legal) requirements.
Thanks for your answer.
I’m not using archetypes for GUI definitions, but I thought adding new readings to observations could only be done by versioning a composition they were a part-of.
This, I believe would force us to create compositions for everything.
The other issues which led us to my previous question, are more of a clinical or medico-legal origin,
I thought that each column in the physiological control form should be a composition, because it needed to be versionable (in order to audit changes) and should be “signed” by the nurse, but (now I see) this does not mean it is a “signed document” as you state for compositions.
Thanks for your answer.
I'm not using archetypes for GUI definitions, but I thought adding new readings to observations could only be done by versioning a composition they were a part-of.
This, I believe would force us to create compositions for everything.
the Composition is indeed the "commit container" but archetypes and content modification is done at any level of granularity. So clinical archetypes typically are defined for the Entry level, i.e. for subtypes. If you are modifying an existing Composition, fine-grained archetypes are used at runtime to control the content modification.
The other issues which led us to my previous question, are more of a clinical or medico-legal origin,
I thought that each column in the physiological control form should be a composition, because it needed to be versionable (in order to audit changes) and should be "signed" by the nurse, but (now I see) this does not mean it is a "signed document" as you state for compositions.