It is hard to get your head around but fun! :)
The current way to design systems is to try to understand the whole of
clinical medicine and design a database schema to match it so that every
piece of data has a matching table or set of fields. This leads to ever
expanding and inflexible schemas that have to be added to or changed
whenever our clinical understanding grows or changes. It means that we as
clinicians have to try to describe clinical concepts to programmers so that
they can build systems that work.
openEHR separates the data model from the semantic model. Lets look at some
xml of a serialised openEHR composition -
(serialised using the .net kernel that is very functional now)
<Item xmlns="openEHR/v1/Composition"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"; xsi:type="section"
archetype_id="openEHR-EHR-SECTION.summary.v1" rm_version="1.0">
<NodeID>at0000</NodeID>
<Item xmlns="openEHR/v1/Composition"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"; xsi:type="evaluation"
archetype_id="openEHR-EHR-EVALUATION.problem-diagnosis-histological.v1"
rm_version="1.0" local_impl_IsSelf="true">
<NodeID>at0000.1.1</NodeID>
<Name xsi:type="dv_coded_text">
<value>Histological diagnosis as defined by a
clinician</value>
<DefiningCode>
<CodeString>at0000.1.1</CodeString>
<TerminologyID>
<ID>local</ID>
</TerminologyID>
</DefiningCode>
</Name>
<Data xsi:type="item_tree" node_id="at0001">
<Item xsi:type="element"
node_id="at0002.1.1">
<Name xsi:type="dv_text">
<value>Histological
diagnosis</value>
</Name>
<Value xsi:type="dv_coded_text">
<value>Seminoma,
testicle</value>
<DefiningCode>
<CodeString>314567</CodeString>
<TerminologyID>
<ID>SNOMED-CT</ID>
</TerminologyID>
</DefiningCode>
</Value>
</Item> <!-- at0002.1.1 -->
</Data>
</Item>
</Item>
This is a subset of data from a sample histology report and is describing
the diagnosis. What you are looking at is data stored using the reference
model. If you look at the data you will see that it is stamped with the
archetype id all the way through and all the nodes are describing very
generic structures. When retrieving this data, the kernel will match it
with the the archetype to validate it. Any archetyped data can be saved
with the same set of reference model classes. So you don't need to keep
changing the database schema to add new clinical concepts, just the
archetypes.
Its worth reading through the openEHR documentation as while there is a lot
of it, it makes the whole concept clearer.
Regards Hugh
-----Original Message-----
From: address@hidden [mailto:address@hidden
On Behalf Of Richard Hosking
Sent: Tuesday, 3 January 2006 9:39 PM
To: General Practice Computing Group Talk
Subject: Re: GP Requirements - was [GPCG_TALK] Re: The Dreaming
I still struggle with this stuff - isnt an archetype describing constraints
on the data- ie type, bounds etc? Shouldnt this be coded into the DB? If not
where?
Richard
Hugh Leslie wrote:
An openEHR database schema doesn't contain ANY archetypes - The data
reflects the openEHR reference model, not the archetype model. If you
have to hard code archetypes in a schema then you have not gained
anything at all. The archetypes just describe what the data means.
_______________________________________________
Gpcg_talk mailing list
address@hidden
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk
_______________________________________________
Gpcg_talk mailing list
address@hidden
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk