Dear John,
You wrote:
Rich says "That kind
of interoperability problem is much more a problem of data variability and
complexity, not a problem of the ontology being “wrong”, but of
there being TOO MUCH DEPTH of ontology challenging the providers."
Are you referring to the classes, or the properties, or both?
Neither
one. I am referring to the diversity of values which can be entered and stored
in response to a question. I.e., if the provider is asked by the EHR what
to put into a cell on the screen, and the answer cannot be predicted as one of
a small number of choices, the entries that are provided vary widely in
practice.
There is an example in the eBook where providers can
enter “shortness of breath”, “difficulty breathing”, or
“SOB”, or other way of recording his observation of the patient
symptoms. Also the diagnosis cells in EHRs require unstructured text to
put in the answers.
The docs using these products are very irritated by
the delays they encounter in “fixing” the entries they made to
within the data constraints which the cells require for full validation to the entire
EHR data structuring requirements.
IMHO the primary issue
with ontologies such as you describe is the sheer number of (class-specifiic)
properties, not so much the depth of the ontology -- users don't interact with
depth directly (or at all), they interact primarily with ontology properties.
The
users are faced with question sequences that make up a tree, depending on which
answer sequence said users enter. That makes a very large set of possible
paths the user can be experiencing through the EHR question sequence.
Some very smart people designed those sequences so that classes and properties
are interpreted (by the EHR) in a programmed way. That makes one hell of
a lot of complexity for the provider to go through.
And
since we have just discussed how every user varies in her word usage,
definitions, applicable rules and so forth, people are being forced into a
single channel of interaction that doesn’t allow the human intuitions to
exercise their usual heuristic efficiencies.
I.e., it
takes much too much time for the docs to deal with the EHR instead of with the
patients.
When you have people
replicating substantial portions of an ontology's class structure as a property
structure, this is where IMHO the idiocy begins. Eliminate that replication,
and the ontology becomes more practical.
John
Yes, and the interaction becomes
unworkable from the user’s view.
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2