I think this statememt of Thomas Beale's is crucial to the understanding..
"But note: EHR systems are not passing around definitions of BP, they are passing around actual BP measurements, which is why archetypes correspond to this interoperability requirement, not underlying ontologies." When health event information is stored in an EHR, there needs to be metadata about a particular concept, not merely the fact that an instance of a concept has occured. I.e. the metadata qualifies the instance. For blood pressure to be recorded, it is common practice to record whether the patient is sitting or standing at the time of measurement. Other metadata that might qualifier the measurement are such things as the instrument used to take the measurement. For a diagnosis of colon cancer, the severity might be recorded using the Dukes Classification. If my understanding is correct, archetypes give a way to define the set of metadata to be associated with a concept, that when applied to an individual instance, capture the nature of the instance sufficiently for it to be understood and communicated clearly and unambiguously by the community of users in the domain. The fundamental elements of ontologies are usually simple - facts, concepts, term definitions. Knowledge is represented by the meaning attached to the definitions, together with the relationship between the terms. The fundamental elements of EHR transactions are complex - instances of archetypes. In order to more clearly discuss these concepts, it would be nice if someone could come up with a name for "instance of archetype" (perhaps something like "archeobject", or "EHRfactlet" or "EHRlet"), to cover the elements of an EHR transaction (diagnostic procedure, diagnosis, therapeutic procedure, adverse health events etc. ). Perhaps Thomas et al already have? eric browne --------------------------------------------------------------- [EMAIL PROTECTED] http://www.healthbase.info ---------------------------------------------------------------
