The General problem with the at codes is that each archetype has the same at codes. Hence it is not an ontology it refers to but is an internal micro-ontology only .
In the DCM approach each node SHALL have a minimum of one external code, preferable Snomed CT, which links the data element in the archetype to an external ontology, which importantly allows external maintenance and governance and facilitates the use in other archetypes or templates as defined in OpenEHR.
That road is seductive but assumes that SNOmed or the like covers the concept unambiguously. The fact is that the world experts on SNOMED spent weeks discussing what codes applied to the BP archetype nodes…and finally agreed with the ones I had proposed.
The fact is that the archetypes are often far less ambiguous than the terms in an ontology.
Modeling in openEHR will accelerate now we have a widening implementation base. The models won’t be perfect or all encompassing but they will support high fidelity processes and sharing evolving data expressions.
Ontological based approaches may be suitable in 20 years, but will need to be based on the clinical models, not the other way around. Clinicians need to define what they want to record.
My 2p - Cheers Sam
Dr Sam Heard
FRACGP, MRCGP, DRCOG, FACHI
Chairman, Ocean Informatics
Chairman, openEHR Foundation
Chairman, NTGPE
+61417838808
The whole point of at codes is that they ARE an internal terminology and that an archetype only has to be consistent internally. Of course any node in an archetype can be bound or mapped to one or more terminologies as necessary. The openEHR work is well proven in many contexts, and your statements are only a matter of opinion rather than any science.