Archetype Design Patterns

At the beginning of 2017 there is no manual for archetype design. Yet we now have nearly a decade of experience in designing and building archetypes.

Will this be possible in the future? Sorry, but probably not.

It is not unreasonable to think that after all this time we should have a huge number of mature archetypes and clear rules that can be applied for any situation. This makes sense from a technical or engineering approach. It does not take into account the messiness and complexity that needs to be represented in medicine.

Anyone can build an archetype.

If it is intended for use in a local system then its design is effectively irrelevant. Just do it!

However if we want a pool of archetypes that will support broad interoperability and sharing of health data - ranging from the simplest to the most complex concepts, from the most general to the most specialised, from single use patterns to fractal data patterns - then we need careful design, tested in implementations. We also need to ensure as best we can that we minimise overlap between concepts as well as minimise gaps.

In practice it has proved harder than it first appears.

Central to design is that clinicians need to drive the requirements, and also the design. It has been observed on many occasions that even with identical clinical requirements, a clinical modeller and a technical modeller will build quite different archetypes. Which archetype best represents the data recording requirements? In most cases it will be the clinical modeller's attempt which is more useful, although technical input will be required to ensure it will be implementable.

This is a useful lesson - clinicians who understand how the data will be captured, used, shared, queried etc will likely create a better archetype. 

Documenting Useful Patterns

But what we can do now is to share some of the design patterns that are proving useful. These patterns are being used as the basis for the next archetype which will need to incorporate additional requirements. This new archetype will no doubt be used as a pattern for future archetypes.

In reality, this is what we do - this is the art of archetyping.

We'd all feel much more comfortable if we could provide foolproof formulas or recipes or rules. If this was possible then it would have been documented and completed long ago.

What will gradually evolve on this page and those that follow will be a synthesis of the experience and knowledge of leading international clinical modellers who are responsible for the archetypes in the international openEHR community's CKM.

We hope it is helpful and that it will grow into a useful resource to the broader openEHR and modelling community.