Hi all,
In the paper “Archetype 101” by Heather and Sam, there’s a short discussion about CDSS, here comes a quote from it:
“Intelligent generic decision support programs can rely on archetype-triggered events and for the
first time operate in real-time.”
I wonder what is the exact meaning of Archetype-triggered events.
IS any research done in this area before or has any implementation based on that been used in a CDSS so far?
Archetypes allow for embedded First Order Predicate Logic therefore in
an application they can do some simple internal processing.
The Epidemiological Surveillance Support System (EpiS3)[1] has been
funded for development (for 3 years, by the INCT-Macc network in Brazil)
based on the Python implementation of openEHR; aka. OSHIP[2].
Though background work began last year, the funding is expected to be
released within the next few weeks and development in earnest can
finally begin. The OSHIP framework also includes event triggering
and we will be exploring several options using both of these triggering
mechanisms in conjunction forward-chaining inference and Bayesian DSS.
At the Zilics EHR there is an implementation of an event service. When a
composition is saved in the database, registered events are triggered.
This is used for several decision support activities. For example, to
help patient triage, an event is triggered when a composition containing
archetypes with information used to decide if the patient situation is
an emergency or not is saved. Then, decision rules are fired and this
will be used to move the patient up or down in the waiting line. To
retrieve the data from the composition when the event is triggered, the
A-Path language is used
(http://www.openehr.org/wiki/display/spec/A-path+proposal+(Zilics)).
We were really saying that it is possible to write AQL statements (or Xpath) against archetypes to test if the last bp.systolic > 140 mmHg for instance. This is due to the fact that the archetype makes it quite straightforward to ask such questions. A number of recent posts have described what is going on in different developments.
Hi Tim,
Thank you for your explanation, may I know more specifically where in archetypes one can put FOL predicates? Should it be added to the XML (or ADL) file , I mean hard coded, or there’s a specific node in archetypes for that, or maybe for each item we may have some predicates…
Is it something done before or in progress?
There’s one section in archetypes named Event which I can see using Ocean Arch. Editor, but I don’t think that’s what you mean.
Hi Tim,
Thank you for your explanation, may I know more specifically where in
archetypes one can put FOL predicates? Should it be added to the XML
(or ADL) file , I mean hard coded, or there's a specific node in
archetypes for that, or maybe for each item we may have some predicates...
Is it something done before or in progress?
There's one section in archetypes named Event which I can see using
Ocean Arch. Editor, but I don't think that's what you mean.
Currently the Archetype Editor does not yet implement invariants. You
would currently have to author them by hand. In ADL 1.4, they go in the
'invariant' section; the proposal in ADL 1.5 is a more powerful syntax
and model, and it would go under a 'rules' section in the archetype -
see http://www.openehr.org/wiki/display/spec/openEHR+Templates+and+Specialised+Archetypes