2014-08-11_Minutes_-_Observables_MO

Meeting Minutes (DRAFT)
Meeting date: August 11th,2014
Group: Observables and Investigation
Model Project
Meeting time: 20:00 – 21:00 UTC
Meeting place: Teleconference
Attendees: Daniel Karlsson (DK), Michael Osborne (MO), David Sperzal (DS), Rob
Hausam (RH), Suzanne Santamaria (SSM), Kent Spackman (KS)
Apologies: nil
Date of next meeting: Monday August 25th
2014
Time of next meeting: 20:00-21:00 UTC
Place of next meeting: Teleconference
DK> I had feedback on the susceptibility testing work but not sample observables. I will be
trying to lure some others in the iPALM SIG to look at the sample observables.
2014-8-11:1 Function Observables
KS> I can make a case for roles “externally grounded dependent entities” (such as “roles”)
and “internally grounded realizable” – It’s the teleology that gets in the way.
DK> The “realizable” are useful in some context. There is a book on functions in the
discussion page for this. There are a lot of different views on functions. I am quite happy
with the way we were talking about representing functions :e.g. ability to hear : property type
person, ‘has realizable’ hearing etc…
CC> I agree, they shouldn’t be “functions”. We are getting closer to the mainstream view of
analytic philosophy.
DK> In conclusion, we will continue on this route, starting with the ICF then nursing SIG
observables.
DS> Should we put together a list of PROPERTY TYPES, to make sure it is workable.
DK> This will be validated in work that will happen in the fall. The VA LEGO work could feed
in.
DS> Yes, we can send you a lot of LEGOs.
DK> The property types will need to be extended, most will not be measurement properties.
We could start with a Wiki Page to discuss and list property types and provide definitions
and examples. Action: DK
2014-8-11:2 Presence observables revisited
DS> The LEGO forms use IS ABOUT with Independent continuant
{Disorders/Procedure/Condition}. We use “Present/Absent/Detected/Not detected” as
answers. LEGO is a question and answer pair. The expressions are compatible with the
observables model.
KS> In some of the more recent diagrams about the explicit context model. There is IS
ABOUT patient life phase, and then that patientLifePhase has an “includes” to a
-
1-
structure/process/disposition, or “does not include” relation when some structure etc doesn’t
exist.
A patient life phase in which something is present is an occurrent, not an independent
continuant. DS>There is a time element involved. KS> Yes. But the LEGO approach being
described seems to dovetail with our revised view of the situation-with-explict-context
hierarchy.
KS> We agree with others who think it should be revised. Particularly when it comes to
presence/absence. “Head injury without loss of consciousness” becomes problematic when
you use presence/absence. People want to make assertions about a patient life phase. A
patient life phase includes structures/processes/dispositions; and/or does not include other
structures/processes/dispositions
DS> “Head injury without loss of consciousness” is this 2 instances?
KS> “Head injury without loss of consciousness” is one observation result, one can create
an Information content entity that is about that PatientLifePhase, which includes “head injury
(a structure)” and does not include “loss of consciousness (a process)”.
KS> Current explicit context model for procedures has good values sets for procedure
contexts, things like planned, done, completed…..
DS> Discernable is meant to be sent to a classifier. But you can’t classify it until procedures
are in the observables model.
KS> LOINC technology preview will be out RSN (real soon now). The workbench for
January 2014 release has object properties needed for this. We will have a top level
concept for observation results, so you can run classifier against your extension. If there are
missing properties that you need to run your classifier, there are ways to patch it with OWL.
2014-8-11:3 Susceptibility Observables
DK> Thanks for comments from MO, JC and RH. I have added a comment about not
reporting the diagnostic antibiotics
Example:
Need to be careful with towards it can be overused.
Then there are mixtures of drugs. E.g. Amoxycillin+clavulanic acid which is a product. There
is no substance for the combination. In order to classify it needs a sub-property chain.
-
2-
KS> In the new substance model we propose to use has_part, e.g. has_part clavulanate
and has_part amoxicillin.
DK> Combinations are synergistic i.e. not just the sum of the parts.
Is it still correct to say the observables subsumes?
KS> An observable about amoxicillin should not subsume an observable about mixtures
that contain amoxicillin.
In substances, we would clearly differentiate a mixture from a non-mixture.
Amoxycillin+Clavulanate is something we would create as a substance. There are two
partitions in the mixture.
DK> This largely avoids this problem. Mixtures could then be represented as substances.
KS> It raises the question of boundary e.g. Multi-Vitamin has a lot of parts of the mixture.
So the boundary rule says we wouldn’t create the mixture unless it is useful to model, for
the therapeutic/biological characteristics (emergent characteristics of the mixture not wholly
found in its partitions)
DK> Will there be a way to represent substances that only contain amoxicillin.
KS> It can be a partition of a mixture or a single substance. E.g. Milk has milk fat, casein,
water etc…
So can we have amoxicillin that is not in a mixture? Every amount of substance that is
defined as “amoxicillin (substance)” will have some impurities and excipients also as part –
so no I don’t think we will have a meaning that represents some quantity of matter that is
“pure amoxicillin”. BUT instead we will have a concept that represents some amount of
matter which is defined by its granular parts being amoxicillin. And a mixture of amoxicillin
and clavulanate will have a has-part relationship to this amoxicillin (substance) which is one
partition. At least that is one way we are proposing. We need to test the model.
DK> I’ll take a look at the substance model and do some more modelling.
KS> Version 0.6 is the most recent version on the collabnet site.
-
3-