Latest round of changes to CPR ontology

[Chimezie Ogbuji]

I am about to commit changes to the CPR ontology (I can't commit now
due to Google Code being in readonly mode now).

Most of the changes were as a result of feedback from Michel Dumontier
(in particular regarding signs and symptoms) from an effort to
integrate the CPR ontology with the Translational Medicine Ontology
[1]. At the same time, a thorough review was done resulting in many
structural changes and the fleshing out of human-readable definitions,
comments, etc.

# Major changes #
## Namespace and organization##
The namespace has been changes to remove reference to a version number
(http://purl.org/cpr/). This new base URI redirects to
http://cpr-ontology.googlecode.com/svn/trunk/CPR.owl. The pheno
namespace has been removed and the terms in that namespace have been
moved under the cpr namespace.

CPR has been separated from BioTop. There is a new file CPR-
biotop.owl which imports CPR and the BioTop-RO-BFO bridge and defines
correspondence between CPR and BioTop.

## Measurements, units and assays ##
The Measurement Units Ontology (MUO [2]) is now imported for use with
assay results (as qualities values with units). The intension is to
capture assay results in one of two ways:

[ a cpr:laboratory-test;
ro:has_participant
[ a cpr:patient
:diastolic [ muo:measuredIn :mmHg;
muo:numericalValue 74 ];
:systolic [ muo:measuredIn :mmHg;
muo:numericalValue 117 ]
]
]

or

[ a cpr:laboratory-test-finding;
ro:has_proper_part
[ a cpr:laboratory-test-finding;
rdfs:label "diastolic BP";
cpr:measurementOf [ muo:measuredIn :mmHg;
muo:numericalValue 74 ]],
[ a cpr:laboratory-test-finding;
rdfs:label "systolic BP";
cpr:measurementOf [ muo:measuredIn :mmHg;
muo:numericalValue 117 ]]
]

Where (in the first case) :diastolic and :systolic are specializations
of muo:qualityValue that relate a participant in an assay with the
measurement of a property of the participant (as a muo:QualityValue).
In the second case, we use cpr:measurementOf to relate the diastolic
BP finding to the diastolic quality value and define the systolic BP
in a similar fashion. cpr:measurementOf is a specialization of
cpr:representationOf that holds between a representational artifact
and a muo:QualityValue (typically associated with the subject of the
clinical finding).

## Symptoms and signs are captured as representational artifacts only
##

The symptom and signs concepts have been modified so that they
primarily are represented as representational artifacts and not
special kinds of bodily features. It is generally a bad idea to mix
the ontology-epistemology divide [3]. The former is about collections
of things in reality and the latter is a study of how cognitive
subjects come to know the truth about phenomena in reality. Some
bodily features, can exist before they are perceived by a patient (who
then hypothesizes they indicates a disease) but they aren't considered
symptoms at that time. So, there is nothing about a cough (for
example) that makes it a symptom besides the way it is reported; this
is an epistemic distinction alone. This is mostly the case with signs
as well although what makes things a sign is a little more objective
or empirical. Still the distinction between signs and other bodily
features is mostly epistemic (whether it is observable or whether it
was observed).

So, now CPR only has symptom-recording and sign-recording. A new
processual entity called self-examination has been added that is an
agentive action performed by a patient on his or herself to determine
the existence of a medical problem. a symptom-recording is then a
primitive clinical-artifact that is the output of a medical-history-
screening-act or self-examination and is the presentation of a bodily
feature. So, here the logical distinction between a symptom and other
bodily features is that they are cpr:representedBy a symptom-recording
rather than some yet to be determined ontologic distinction. As a
result, a symptom-recording is no longer a clinical-finding since it
could be the output of a self-examination and thus not composed by a
clinician.

Similarly, a sign-recording is a clinical-finding that is the output
of a clinical-examination. So, a sign can be logically distinguished
from other bodily features by the fact that they are cpr:representedBy
a sign-recording. However, as is the case with symptoms, we don't
make an explicit sign class due to the problems mentioned before about
the epistemic nature of the distinction from other bodily features.

## Uncertain temporal extent of occurrents ##

4 new datatype predicates were added as specializations of
cpr:occursDuring that relate a processual entity with a temporal
region (during which they occur) where there is uncertainty about the
specific begin and end: cpr:startsNoEarlierThan,
cpr:startsNoLaterThan, cpr:stopsNoEarlierThan, and
cpr:stopsNoLaterThan. The range of these predicates is xsd:dateTime.

## Complications and sequela ##
A new cpr:sequela class has been added, a defined class that is those
bodily features that are the consequence of a disease

# Minor changes #

cpr:SurgicalMethod is no longer camel-case (which differed from the
common naming convention) and is a processual entity. cpr:anatomical-
structure is a snap:Object. cpr:clinical-act is now an agentive action
(it doesn't just have a clinician as a participant but instead
ro:has_agent is used). A cpr:clinical-investigation-act now
cpr:investigates an etiologic-agent or something that indicates a
particular cpr:therapeutic-act. cpr:dependent-organismal-continuant
has been removed as an intermediary class between
snap:DependentContinuant and immaterial-pathological-entities, spaces,
etc. cpr:deoxyribonucleic-acid-structure has been added (used for he
definition of cpr:genetic-abnormality). cpr:material-anatomical-
entity has been removed.

There have been several clarifications to the disease pathology
model. Per whitbecks characterization of morphologic alterations as a
component of a disease (pathological), cpr:material-pathological-
entity is now defined as a transformation of an cpr:anatomical-
structure. This is inline with SNOMED-CTs characterization of the
ASSOCIATED MORPHOLOGY attribute. cpr:material-pathological-entity is
also defined to cpr:participates_in in a cpr:morphologic-alteration.
cpr:morphologic-alteration is a ro:part_of a cpr:pathological-process
and has (as participant) a cpr:physical-anatomical-entity (which can
be transformed into a cpr:material-pathological-entity). A disease is
now the agent of a pathological process that has (as a component) a
cpr:morphologic-alteration. This still leaves out disease processes
involving physiologic alterations (as described by Whitbeck) rather
than physical alterations. This needs to be addressed.

The cpr:medical-problem class now includes things represented by sign-
recordings and symptom-recordings. cpr:organism has been added (of
which a pathogen is now a specialization of ). cpr:biological-process
has been added (from OBR) that are one of the 3 components of bodily
features (from OGMS). cpr:healthcare-professional-role has been
added.

cpr:perceiver-role and cpr:sensation have been removed (now that
symptom-recordings are defined via the new cpr:self-examination
class). A vital sign is now *not* a sign but a special kind of bodily
feature. The suffix was a misnomer previously.

## Changes to relations ##

The domain and range of various 'core' relations has been added (such
as cpr:hasInput).

Various relations were removed due to their idiosyncratic nature:
- cpr:hasComplication / cpr:isComplicationOf was removed due to
introduction of cpr:sequela which can be used with cpr:hasConsequence
for the same purpose
- cpr:hasImmediateConsequence and cpr:hasLateConsequence were removed
(better served via use of occurrents with uncertain temporal extents)
- cpr:isTreatedBy

The domain of both cpr:hasIndication and cpr:hasContraindication is
therapeutic-act. The range of cpr:hypothesizedProblem (whose domain is
cpr:clinical-diagonosis) is now disease rather than medical problem.
The intuition here is that (in the medical literature) a diagnosis is
a conclusion about a disease specifically.

[1] http://code.google.com/p/translationalmedicineontology/
[2] http://forge.morfeo-project.org/wiki_en/index.php/Units_of_measurement_ontology
[3] Bodenreider O, Smith B, Burgun A, The Ontology-Epistemology
Divide: A Case Study in Medical Terminology. In: Varzi AC and Vieu L,
editors. FOIS 2004. Proceedings of. The International Conferenceon
Formal Ontology and Information Systems; 2004 Nov 4-6, Turin, Italy.
Amsterdam IOS Press; 2004.185–195

[Michel Dumontier]

Chimezie,

  good job. comments inline.

m.

On Wed, Jun 9, 2010 at 11:02 AM, Chimezie Ogbuji <chim...@gmail.com> wrote:

I am about to commit changes to the CPR ontology (I can't commit now
due to Google Code being in readonly mode now).

Most of the changes were as a result of feedback from Michel Dumontier
(in particular regarding signs and symptoms) from an effort to
integrate the CPR ontology with the Translational Medicine Ontology
[1].  At the same time, a thorough review was done resulting in many
structural changes and the fleshing out of human-readable definitions,
comments, etc.

# Major changes #
## Namespace and organization##
The namespace has been changes to remove reference to a version number
(http://purl.org/cpr/).  This new base URI redirects to
http://cpr-ontology.googlecode.com/svn/trunk/CPR.owl.  The pheno
namespace has been removed and the terms in that namespace have been
moved under the cpr namespace.

CPR has been separated from BioTop.  There is a new file CPR-
biotop.owl which imports CPR and the BioTop-RO-BFO bridge and defines
correspondence between CPR and BioTop.

## Measurements, units and assays ##
The Measurement Units Ontology (MUO [2]) is now imported for use with
assay results (as qualities values with units).

I think the only thing they do wrong is that quality types (e.g. mass) are represented as instances.

 The intension is to
capture assay results in one of two ways:

[ a cpr:laboratory-test;
 ro:has_participant
   [ a cpr:patient
     :diastolic [ muo:measuredIn :mmHg;
                  muo:numericalValue 74 ];
     :systolic  [ muo:measuredIn :mmHg;
                  muo:numericalValue 117 ]
   ]
]

i don't like this because

1 - the patient should have the quality of diastolic/systolic blood pressure, and that the measured values should be then specified  against these quality instances (as opposed to using these as predicates)

2 - since these values change with time, they need to be directly linked to the test, rather than through the patient 

 

or

[ a cpr:laboratory-test-finding;
 ro:has_proper_part
    [ a cpr:laboratory-test-finding;
      rdfs:label "diastolic BP";
      cpr:measurementOf [ muo:measuredIn :mmHg;
                          muo:numericalValue 74 ]],
    [ a cpr:laboratory-test-finding;
      rdfs:label "systolic BP";
      cpr:measurementOf [ muo:measuredIn :mmHg;
                          muo:numericalValue 117 ]]
]

a test finding should be composed of the informational elements - e.g. the values themselves, and these values should then be linked to the quality they measure, and that quality to the patient.

 

Where (in the first case) :diastolic and :systolic are specializations
of muo:qualityValue that relate a participant in an assay with the
measurement of a property of the participant (as a muo:QualityValue).
In the second case, we use cpr:measurementOf to relate the diastolic
BP finding to the diastolic quality value and define the systolic BP
in a similar fashion.  cpr:measurementOf is a specialization of
cpr:representationOf that holds between a representational artifact
and a muo:QualityValue (typically associated with the subject of the
clinical finding).

## Symptoms and signs are captured as representational artifacts only
##

The symptom and signs concepts have been modified so that they
primarily are represented as representational artifacts and not
special kinds of bodily features.  It is generally a bad idea to mix
the ontology-epistemology divide [3].  The former is about collections
of things in reality and the latter is a study of how cognitive
subjects come to know the truth about phenomena in reality.  Some
bodily features, can exist before they are perceived by a patient (who
then hypothesizes they indicates a disease) but they aren't considered
symptoms at that time.  

some? surely all bodily features exist independent of whether we perceive them.

 

So, there is nothing about a cough (for
example) that makes it a symptom besides the way it is reported; this
is an epistemic distinction alone.  This is mostly the case with signs
as well although what makes things a sign is a little more objective
or empirical.  Still the distinction between signs and other bodily
features is mostly epistemic (whether it is observable or whether it
was observed).

yes - signs and symptoms are cognitive associations or interpretations of bodily feature(s) as evidence for or against the belief that some condition exists.

 

So, now CPR only has symptom-recording and sign-recording.  A new
processual entity called self-examination has been added that is an
agentive action performed by a patient on his or herself to determine
the existence of a medical problem.  a symptom-recording is then a
primitive clinical-artifact that is the output of a medical-history-
screening-act or self-examination and is the presentation of a bodily
feature.  So, here the logical distinction between a symptom and other
bodily features is that they are cpr:representedBy a symptom-recording
rather than some yet to be determined ontologic distinction.  As a
result, a symptom-recording is no longer a clinical-finding since it
could be the output of a self-examination and thus not composed by a
clinician.

hmm... i don't agree. Surely we can instantiate signs and symptoms directly with an indication of the bodily features from which they the cognitive associations are made, and indicate who is responsible for them. in the worst case, these bodily features play the role of signs/symptoms depending on whom is involved in the diagnosing process - although whether this is a role in the sense of BFO is unclear to me.

--
Michel Dumontier
Associate Professor of Bioinformatics
Carleton University
http://dumontierlab.com

[Chimezie Ogbuji]

I have committed the changes.

-- Chime

On Jun 9, 11:02 am, Chimezie Ogbuji <chime...@gmail.com> wrote:
> I am about to commit changes to the CPR ontology (I can't commit now
> due to Google Code being in readonly mode now).
>
> Most of the changes were as a result of feedback from Michel Dumontier
> (in particular regarding signs and symptoms) from an effort to
> integrate the CPR ontology with the Translational Medicine Ontology
> [1].  At the same time, a thorough review was done resulting in many
> structural changes and the fleshing out of human-readable definitions,
> comments, etc.
>
> # Major changes #
> ## Namespace and organization##
> The namespace has been changes to remove reference to a version number

> (http://purl.org/cpr/).  This new base URI redirects tohttp://cpr-ontology.googlecode.com/svn/trunk/CPR.owl.  The pheno

> [2]http://forge.morfeo-project.org/wiki_en/index.php/Units_of_measuremen...

[Sivaram Arabandi]

Nice work Chime. Please see my comments inline ...

Sivaram

Agree with Michel. I much rather prefer the way it is done in PATO and UO  ( http://purl.org/obo/owl/PATO)

 The intension is to
capture assay results in one of two ways:

[ a cpr:laboratory-test;
 ro:has_participant
   [ a cpr:patient
     :diastolic [ muo:measuredIn :mmHg;
                  muo:numericalValue 74 ];
     :systolic  [ muo:measuredIn :mmHg;
                  muo:numericalValue 117 ]
   ]
]

i don't like this because

1 - the patient should have the quality of diastolic/systolic blood pressure, and that the measured values should be then specified  against these quality instances (as opposed to using these as predicates)

Agree. Blood pressure is a bodily feature. Findings refer to bodily features and are associated with a value.
 

2 - since these values change with time, they need to be directly linked to the test, rather than through the patient 

Disagree here. Findings are 'about' a patient. Therefore, should be related to patient. A 'test' is only a means to the Finding. Michel infact says this in his comment below the next example of BP
Minor point - I wouldn't relate blood pressure finding to a laboratory test  :)
 

 

or

[ a cpr:laboratory-test-finding;
 ro:has_proper_part
    [ a cpr:laboratory-test-finding;
      rdfs:label "diastolic BP";
      cpr:measurementOf [ muo:measuredIn :mmHg;
                          muo:numericalValue 74 ]],
    [ a cpr:laboratory-test-finding;
      rdfs:label "systolic BP";
      cpr:measurementOf [ muo:measuredIn :mmHg;
                          muo:numericalValue 117 ]]
]

a test finding should be composed of the informational elements - e.g. the values themselves, and these values should then be linked to the quality they measure, and that quality to the patient.

 

Where (in the first case) :diastolic and :systolic are specializations
of muo:qualityValue that relate a participant in an assay with the
measurement of a property of the participant (as a muo:QualityValue).
In the second case, we use cpr:measurementOf to relate the diastolic
BP finding to the diastolic quality value and define the systolic BP
in a similar fashion.  cpr:measurementOf is a specialization of
cpr:representationOf that holds between a representational artifact
and a muo:QualityValue (typically associated with the subject of the
clinical finding).

This is how I am representing this in the SDO:
DiastolicBloodPressureFinding
       representationOf DiastolicBloodPressure and
       hasValue ""^^xsd:int and
       hasUnit MillimetersOfMercury

  - DiastolicBloodPressure is a bodily feature (defined in SDO)
  - MillimetersOfMercury is one of the many units to measure pressure (defined in SDO-UnitsOntology)
 

Not sure of all the distinctions described above (and below). From a clinical point of view:
     - Symptoms (and Signs) are bodily features or changes in bodily features reported by a patient (and observed by a Physician respectively). In this sense, I see them as defined classes.
     - They exist whether they are recorded or not.
     - A clinical history procedure results in a list of zero or more symptoms
     - A physical exam procedure results in a list of zero or more signs
                  - so I am not sure what the difference is between a sign and sign-recording!

 

Similarly, a sign-recording is a clinical-finding that is the output
of a clinical-examination.  So, a sign can be logically distinguished
from other bodily features by the fact that they are cpr:representedBy
a sign-recording.  However, as is the case with symptoms, we don't
make an explicit sign class due to the problems mentioned before about
the epistemic nature of the distinction from other bodily features.

## Uncertain temporal extent of occurrents ##

4 new datatype predicates were added as specializations of
cpr:occursDuring that relate a processual entity with a temporal
region (during which they occur) where there is uncertainty about the
specific begin and end: cpr:startsNoEarlierThan,
cpr:startsNoLaterThan, cpr:stopsNoEarlierThan, and
cpr:stopsNoLaterThan. The range of these predicates is xsd:dateTime.

Like this.
 

## Complications and sequela ##
A new cpr:sequela class has been added, a defined class that is those
bodily features that are the consequence of a disease

Did you add both terms?
The causality can be broadened to include not only disease but also procedures and trauma.
 

Disagree. Vital signs comprise a subset of signs which in turn are bodily features.

## Changes to relations ##

The domain and range of various 'core' relations has been added (such
as cpr:hasInput).

Various relations were removed due to their idiosyncratic nature:
- cpr:hasComplication / cpr:isComplicationOf was removed due to
introduction of cpr:sequela which can be used with cpr:hasConsequence
for the same purpose

hmmm. Complication = Sequela ? not so sure. The relation may be more of Sequelae being complications that occur late (how late is very subjective). So if something bad happens during a disease course or after it has subsided (or attained a more chronic phase), or during or immediately following a procedure or trauma, it is called a complication. If something bad happens several months or years later, it is often referred to as sequela.
 

- cpr:hasImmediateConsequence and cpr:hasLateConsequence were removed
(better served via use of occurrents with uncertain temporal extents)
- cpr:isTreatedBy

The domain of both cpr:hasIndication and cpr:hasContraindication is
therapeutic-act. The range of cpr:hypothesizedProblem (whose domain is
cpr:clinical-diagonosis) is now disease rather than medical problem.
The intuition here is that (in the medical literature) a diagnosis is
a conclusion about a disease specifically.

[1] http://code.google.com/p/translationalmedicineontology/
[2] http://forge.morfeo-project.org/wiki_en/index.php/Units_of_measurement_ontology
[3] Bodenreider O, Smith B, Burgun A, The Ontology-Epistemology
Divide: A Case Study in Medical Terminology. In: Varzi AC and Vieu L,
editors. FOIS 2004. Proceedings of. The International Conferenceon
Formal Ontology and Information Systems; 2004 Nov 4-6, Turin, Italy.
Amsterdam IOS Press; 2004.185–195

--
Michel Dumontier
Associate Professor of Bioinformatics
Carleton University
http://dumontierlab.com

Sivaram Arabandi

[Chimezie Ogbuji]

Hey Michel.

On Jun 9, 12:30 pm, Michel Dumontier <michel.dumont...@gmail.com>
wrote:

> Chimezie,
>   good job. comments inline.

Thanks.

> > ## Measurements, units and assays ##
> > The Measurement Units Ontology (MUO [2]) is now imported for use with
> > assay results (as qualities values with units).
> I think the only thing they do wrong is that quality types (e.g. mass) are
> represented as instances.

Yes, I found it a bit odd when I noticed that. However, are you
referring to what they call 'units' (see below)?

>  The intension is to> capture assay results in one of two ways:
>
> > [ a cpr:laboratory-test;
> >  ro:has_participant
> >    [ a cpr:patient
> >      :diastolic [ muo:measuredIn :mmHg;
> >                   muo:numericalValue 74 ];
> >      :systolic  [ muo:measuredIn :mmHg;
> >                   muo:numericalValue 117 ]
> >    ]
> > ]
>
> i don't like this because
> 1 - the patient should have the quality of diastolic/systolic blood
> pressure, and that the measured values should be then specified  against
> these quality instances (as opposed to using these as predicates)

But I think this is what the statements above intend to represent.
Since :diastolic is a rdfs:subProperty of muo:qualityValue it captures
'the quality value of a measurable physical quality of an entity or
phenomenon.' In this case the entity is the patient and the object is
the the value of an individual quality (the value of the diastolic BP
quality at the time of measurement - although as you point out later
the relationship between the quality value and when it is measured is
disconnected).

MUO includes a muo:measuresQuality relationship that holds between a
unit of measurement and a muo:PhysicalQuality (snap:Quality), so to
make an explicit statement about the BP pressure quality you could do:

:mmHg
muo:measuresQuality :DiastolicBloodPressure, :SystolicBloodPressure
:DiastolicBloodPressure a snap:Quality
:SystolicBloodPressure a snap:Quality

So, if you preferred to not use the relation to capture which quality
we are measuring, you could (perhaps) do this instead:

 [ a cpr:patient
     muo:qualityValue [ muo:measuredIn
[ muo:measuresQuality :DiastolicBloodPressure ];
                  muo:numericalValue 74 ];
 muo:qualityValue  [ muo:measuredIn
[ muo:measuresQuality :SystolicBloodPressure ];
muo:numericalValue 117 ]]

I have yet to fully understand the various ways you can capture
qualities, their values, and units via the MUO, but I get the
impression that it is quite versatile and supports various alternative
ways to do so (in a way that is still consistent with the BFO notions
of inherence, qualitis, etc.)

> 2 - since these values change with time, they need to be directly linked to
> the test, rather than through the patient

Agreed, which is the reason why I personally prefer the second
approach.

> > [ a cpr:laboratory-test-finding;
> >  ro:has_proper_part
> >     [ a cpr:laboratory-test-finding;
> >       rdfs:label "diastolic BP";
> >       cpr:measurementOf [ muo:measuredIn :mmHg;
> >                           muo:numericalValue 74 ]],
> >     [ a cpr:laboratory-test-finding;
> >       rdfs:label "systolic BP";
> >       cpr:measurementOf [ muo:measuredIn :mmHg;
> >                           muo:numericalValue 117 ]]
> > ]
>
> a test finding should be composed of the informational elements - e.g. the
> values themselves, and these values should then be linked to the quality
> they measure, and that quality to the patient.

So, in the second approach (above), the test finding is 'about' (or a
representation of) quality values that can be linked to qualities (via
muo:measuresQuality) which can then be said to inhere in the subject
of the finding, etc.

>> ..snip ..

> > subjects come to know the truth about phenomena in reality.  Some
> > bodily features, can exist before they are perceived by a patient (who
> > then hypothesizes they indicates a disease) but they aren't considered
> > symptoms at that time.
>
> some? surely all bodily features exist independent of whether we perceive
> them.

Yes, *all* bodily features that are indicated to be 'signs' or
'symptoms' by physician and patient alike exist independent of whether
they are perceived and how they are perceived - which is the problem
of trying to distinguish signs & symptoms from other bodily features
by virtue of how they are perceived (which has nothing to do with
their nature)

> > So, now CPR only has symptom-recording and sign-recording.  A new
> > processual entity called self-examination has been added that is an
> > agentive action performed by a patient on his or herself to determine
> > the existence of a medical problem.  a symptom-recording is then a
> > primitive clinical-artifact that is the output of a medical-history-
> > screening-act or self-examination and is the presentation of a bodily
> > feature.  So, here the logical distinction between a symptom and other
> > bodily features is that they are cpr:representedBy a symptom-recording
> > rather than some yet to be determined ontologic distinction.  As a
> > result, a symptom-recording is no longer a clinical-finding since it
> > could be the output of a self-examination and thus not composed by a
> > clinician.

> hmm... i don't agree. Surely we can instantiate signs and symptoms directly
> with an indication of the bodily features from which they the cognitive
> associations are made, and indicate who is responsible for them.

I'm not sure when you say 'signs' or 'symptoms' whether you are
referring to 1) the digital artifacts in the medical record or 2)
some other thing, i.e., the 'interpretations' of bodily feature(s).
And if you meant 2), what kind of things are they if they are not
digital artifacts or bodily features?

> in the worst case, these bodily features play the role of signs/symptoms depending
> on whom is involved in the diagnosing process - although whether this is a
> role in the sense of BFO is unclear to me.

Ok, so this is what I understood originally from your take on signs
and symptoms. Assuming they are roles (SignRole and SymptomRole for
instance), the bodily features that 'play' these roles in a diagnosing
process are considered signs and symptoms:

Class: Sign
EquivalentTo:
cpr:bodily-feature that (cpr:hasRole some SignRole) and
(ro:participates_in some
(cpr:clinical-examination that (ro:has_participant some
cpr:physician)))

That makes sense for those signs and symptoms that are continuants,
but not for those that are processes

-- Chime

[Michel Dumontier]

the only point here is that the nested structure does not link the data correctly.

a finding 'is about' the bodily feature.

a representation of, is a model, diagram or illustration of. so i find that kinda weird to use it here.

the bodily features do, yes.  sign/symptoms -> without being observed for the purpose of diagnosis? odd.

Agree that vital signs are a kind of sign.

[Michel Dumontier]

exactly. 

MUO includes a muo:measuresQuality relationship that holds between a
unit of measurement and a muo:PhysicalQuality (snap:Quality), so to
make an explicit statement about the BP pressure quality you could do:

:mmHg
muo:measuresQuality :DiastolicBloodPressure, :SystolicBloodPressure
:DiastolicBloodPressure a snap:Quality
:SystolicBloodPressure a snap:Quality

MUO should state an existential restriction,

muo:mmHg a [ 

  a owl:Restriction; 

  owl:onProperty muo:measuresQuality;

  owl:someValuesFrom :DiastolicBloodPressure] .

where as real instances can be linked together, as in:

muo:mmHg muo:measuresQuality :quality1 .

:quality1 a :DiastolicBloodPressure .

 

So, if you preferred to not use the relation to capture which quality
we are measuring, you could (perhaps) do this instead:

        [ a cpr:patient
          muo:qualityValue [ muo:measuredIn
[ muo:measuresQuality :DiastolicBloodPressure ];
                                      muo:numericalValue 74 ];
         muo:qualityValue  [ muo:measuredIn
[ muo:measuresQuality :SystolicBloodPressure ];
                                      muo:numericalValue 117 ]]

umm i don't know but that all seems weird 

in sio - i would do this by 

[] a cpr:patient; 

   'has quality' [

       a :diastolicBloodPressure;

       'has measurement value' [ a 'quantity'; 'has value' "74"; 'has unit' 'mmHg';]]

right, so here's how i would do it in SIO

[] a cpr:laboratory-test-finding;
   'has proper part' [ a 'quantity'; 'has value' "74"; 'has unit' 'mmHg';

                             'is measurement value of' [ a 'diastolic blood pressure';

                                                                    'is quality of' 'patient X']]

   'has proper part' [ a 'quantity'; 'has value' "117"; 'has unit' 'mmHg';

                             'is measurement value of' [ a 'systolic blood pressure';

                                                                    'is quality of' 'patient X']]

 

>> ..snip ..

> > subjects come to know the truth about phenomena in reality.  Some
> > bodily features, can exist before they are perceived by a patient (who
> > then hypothesizes they indicates a disease) but they aren't considered
> > symptoms at that time.
>
> some? surely all bodily features exist independent of whether we perceive
> them.

Yes, *all* bodily features that are indicated to be 'signs' or
'symptoms' by physician and patient alike exist independent of whether
they are perceived and how they are perceived - which is the problem
of trying to distinguish signs & symptoms from other bodily features
by virtue of how they are perceived (which has nothing to do with
their nature)

indeed

 

> > So, now CPR only has symptom-recording and sign-recording.  A new
> > processual entity called self-examination has been added that is an
> > agentive action performed by a patient on his or herself to determine
> > the existence of a medical problem.  a symptom-recording is then a
> > primitive clinical-artifact that is the output of a medical-history-
> > screening-act or self-examination and is the presentation of a bodily
> > feature.  So, here the logical distinction between a symptom and other
> > bodily features is that they are cpr:representedBy a symptom-recording
> > rather than some yet to be determined ontologic distinction.  As a
> > result, a symptom-recording is no longer a clinical-finding since it
> > could be the output of a self-examination and thus not composed by a
> > clinician.

> hmm... i don't agree. Surely we can instantiate signs and symptoms directly
> with an indication of the bodily features from which they the cognitive
> associations are made, and indicate who is responsible for them.

I'm not sure when you say 'signs' or 'symptoms' whether you are
referring to 1) the digital artifacts in the medical record

no 

or 2)
some other thing, i.e., the 'interpretations' of bodily feature(s).

yes 

And if you meant 2), what kind of things are they if they are not
digital artifacts or bodily features?

reading on...

 

> in the worst case, these bodily features play the role of signs/symptoms depending
> on whom is involved in the diagnosing process - although whether this is a
> role in the sense of BFO is unclear to me.

Ok, so this is what I understood originally from your take on signs
and symptoms.  Assuming they are roles (SignRole and SymptomRole for
instance), the bodily features that 'play' these roles in a diagnosing
process are considered signs and symptoms:

Class: Sign
EquivalentTo:
cpr:bodily-feature that (cpr:hasRole some SignRole) and
(ro:participates_in some
 (cpr:clinical-examination that (ro:has_participant some
cpr:physician)))

if we go down the rout of roles:

Sign 

 EquivalentTo:

  cpr:bodily-feature that ('has role' some ('indication role' that 'is realized in' some 'clinical examination' that 'realizes' some ('physician role' that 'is the role of' some person)))

 

That makes sense for those signs and symptoms that are continuants,
but not for those that are processes

lol - so many challenges when using the BFO.. indeed, processes are not (currently) capable of bearing qualities, nor realizables.

i suggest something a bit more abstract -> the generically dependent continuant is the current BFO dustbin for everything else. so, a sign can be generically dependent on instances of any kind (they are types), which i suggest might (subject to independent confirmation) include processes.

in SIO, we would deal with signs/symptoms through more abstract types -> symbols, which agents correctly interpret with the appropriate background knowledge.

m.

 

-- Chime

[Chimezie Ogbuji]

Hey Siva.

On Jun 9, 5:02 pm, Sivaram Arabandi <sivaram.araba...@gmail.com>
wrote:
... snip ..

> > I think the only thing they do wrong is that quality types (e.g. mass) are
> > represented as instances.
>

> Agree with Michel. I much rather prefer the way it is done in PATO and UO  (http://purl.org/obo/owl/PATO) <http://purl.org/obo/owl/PATO>

It is not clear to me in that approach how qualities (which inhere in
a continuant) are related to a scalar value and unit. For example,
how would the blood pressure measurement example be captured in that
framework?

> i don't like this because
> > 1 - the patient should have the quality of diastolic/systolic blood
> > pressure, and that the measured values should be then specified  against
> > these quality instances (as opposed to using these as predicates)
>
> Agree. Blood pressure is a bodily feature. Findings refer to bodily features
> and are associated with a value.

See my response to Michel in my previous email.

>
> > 2 - since these values change with time, they need to be directly linked to
> > the test, rather than through the patient

> Disagree here. Findings are 'about' a patient. Therefore, should be related
> to patient. A 'test' is only a means to the Finding. Michel infact says this
> in his comment below the next example of BP

So, I think the general issue here is how do you give a temporal
context to a particular BP measurement? If I measure 130 over 90 at
10:30am and 112 over 80 at 1:30pm, how do I distinguish both quality
values by the time they were recorded? There are several ways this
can be done in the currently:

I. MUO has a muo:inTime datatype relation that holds between a
muo:QualityValue and an xsd:dateTime which is "The time of a
particular quality value (e.g. the weight of Carlos yesterday)".
II. The provenance associated with the clinical-finding that
represents the measured BP value can be a basis for this (i.e., when
it was entered into the record - which of course could differ from
when the clinical-examination actually happened)
III. The temporal extent of the clinical-examination whose output is
the clinical-finding that represents the measured BP value can be used
as a basis for this (i.e., when was the BP cuff put on the patient by
the nurse / physician)

> Minor point - I wouldn't relate blood pressure finding to a laboratory test
> :)

s/cpr:laboratory-test-finding/cpr:clinical-examination?

> This is how I am representing this in the SDO:
> DiastolicBloodPressureFinding
>        representationOf DiastolicBloodPressure and
>        hasValue ""^^xsd:int and
>        hasUnit MillimetersOfMercury
>
>   - DiastolicBloodPressure is a bodily feature (defined in SDO)
>   - MillimetersOfMercury is one of the many units to measure pressure
> (defined in SDO-UnitsOntology)

Ok.

> Not sure of all the distinctions described above (and below). From a
> clinical point of view:
>      - Symptoms (and Signs) are bodily features or changes in bodily
> features reported by a patient (and observed by a Physician respectively).
> In this sense, I see them as defined classes.
>      - They exist whether they are recorded or not.
>      - A clinical history procedure results in a list of zero or more
> symptoms
>      - A physical exam procedure results in a list of zero or more signs
>                   - so I am not sure what the difference is between a sign
> and sign-recording!

Ok. Using an example from the wikipedia page about medical signs (for
a lack of a better example): Arcus senilis. It is a white or gray
opaque ring in the corneal margin (peripheral corneal opacity). It
can be an indicator of conditions such as hypercholesterolemia. When
I extract Arcus senilis from SNOMED-CT, I get:

Class: sno:ArcusSenilis
SubClassOf:
cpr:pathological-disposition,
( ro:has_part some ( ( sno:findingSite some
sno:PeripheralCornea ) and ( sno:associatedMorphology some
sno:Deposition ) ) ),
( ro:has_part some ( ( sno:associatedMorphology some
sno:Degeneration ) and ( sno:findingSite some
sno:PeripheralCornea ) )

So let me give my standard disclaimer: I've never claimed to be a
physician so don't expect my clinical examples to make perfect sense,
but I hope they capture the general idea :).

Let's assume a physician examines a patient's eyes and observes
sno:ArcusSenilis. After 'converting' this SNOMED-CT definition into
the current framework, we have a degenerated, deposited abnormality at
the sno:PeripheralCornea that is the bodily-feature observed as a sign
(a cpr:material-pathological-entity). Then the physician creates a
cpr:sign-recording entry in the patient's record (i.e., a digital
artifact in the patient record that represents this degenerated,
deposited white or gray opaque ring):

_:recording1 a cpr:sign-recording;
cpr:representationOf _:sign1 ;
dc:date "2008-01-10"^^xsd:date .
_:sign1 a cpr:material-pathological-entity;
ro:located_in [ a sno:PeripheralCornea ];
ro:participates_in [ a sno:Deposition ], [ a
sno:Degeneration ]

So, _:recording1 is the sign-recording (a representational artifact
entered into the patient record on january 10th 2008) and _:sign1 is
what this artifact is/was 'about' ( a degenerated, deposition at the
peripheral cornea )

> ..snip ..

> > ## Complications and sequela ##
> >> A new cpr:sequela class has been added, a defined class that is those
> >> bodily features that are the consequence of a disease
>
> Did you add both terms?

Just sequela. I'm still not clear about the distinction between the
two as clinicians use the terms. Stedman's medical dicitonary says
the following about sequela:

"A condition following as a consequence of a disease."

It also says the following about the term 'complication':

"A morbid process or event that occurs during the course of a disease
that is not an essential part of that disease, although it may result
from it or from independent causes."

> The causality can be broadened to include not only disease but also
> procedures and trauma.

Yes, and the isConsequenceOf relation can be used for this purpose,
since unlike ro:agent_in, its domain and range are not restricted.

..snip..

-- Chime

[Bob Powers]

The ontology-epistemology divide seems to me to be different from the distinction between an entity and the record of it.  (Chime, thanks for the reference to this http://ontology.buffalo.edu/medo/Onto_Epist.pdf.)  Just because an entity has some of the properties of a record does not make it so.  Since I'm of the opinion that signs/symptoms are not records, making a clear distinction between sign and sign-record is very reassuring.

The question is, how should 'sign' be constructed?  I would hope that there would be some formulation for what a (universal) sign is logically prior to its being referred to by a (universal) record.

If I could be permitted to restate the role route so I can see it better:

Sign == 'bodily feature' that

  ('has role' some

    ('indication role' that

      ('is realized in' some

        ('clinical examination' that

          (realizes some

            ('physician role' that

              ('is the role of' some person)))))))

And again, without the class conjunctions and existential restrictions:

'clinical examination'

  realizes 'indication role'

    'is role of' 'bodily feature'

  realizes 'clinician role'

    'is role of' person

So this role route must be rejected because only a continuant can have a role.  As I understand it, Michel's proposal would replace 'bodily feature' above by GDC that inheresIn 'bodily feature':

'clinical examination'

  realizes 'indication role'

    'is role of' GDC

      inheresIn 'bodily feature'

  realizes 'clinician role'

    'is role of' person

Now, my copy of BFO 1.1 downloaded from googlecode does not enjoin one from having a GDC depending on an occurrent.  But tmo-external.owl has a comment for GDC:

"Definition: A continuant [snap:Continuant] that is dependent on one or other independent continuant [snap:IndependentContinuant] bearers. For every instance of A requires some instance of (an independent continuant [snap:IndependentContinuant] type) B but which instance of B serves can change from time to time."^^string

This would seem to restrict GDC's to depending on ICs.  Too bad, I think that we really need signs/symptoms that are able to be (or are able to be closely tied to) occurrents so as not prematurely to model too closely.

So we are trying to cross the ontology-epistemology <chasm>, but with purely BFO-realist-ontology tools that cannot step outside the domain.  This is where, I take it, Michel's SIO solution does indeed step outside.

Dare I suggest an 'epistemically dependent continuant'?  One that inheres in an entity and that has an interpretive role that is realized in an interpretive process.  The chasm is more fundamental than a distinction between continuants and occurrents, or between specific and generic dependency.  But maybe one can just label the thing and move on.  Have no idea whether this is worthy or original, and it's way beyond the purview of the CPR itself.  But at least it's got a target painted on its back:)

My strong prejudice is that, whatever solution emerges, it is worth getting right at some granularity for the sake of the whole stack.

-Bob

 

[Sivaram Arabandi]

If I am not mistaken, there are three things here:

1. a 'thing' in reality  e.g.: 'Blood Pressure'

                                    e.g.: 'Arcus senilis'

2. a 'fact' about the thing in reality   e.g.: Blood Pressure value of 120/80 mm Hg 

                                                               e.g.: Arcus senilis present

3. a 'recording' of the fact  e.g. a paper chart with an 'entry of the fact'

                                               e.g. a digital patient record (EMR...) with an 'entry of the fact'

Then there is other provenance information we can capture about the 'fact' e.g. for Blood Pressure:

- measurement datetime: e.g. at 3:15 pm on 6/10/2010

- body position of the patient:  sitting, standing, supine, ....

- measurement site:  right upper limb, left ankle ....

- (physiological) state: resting, exercising

- measured by: intra-arterial, pressure cuff method.....

- ....

Analyzing this further:

1. 'Blood Pressure' is a bodily feature and a quality of the patient. This is also considered a vital sign.

2. The Blood Pressure value would be a Clinical Finding represented as:

Blood Pressure Finding := 

Situation  that  

includes  Blood Pressure  

and  hasMeasuredValue ""^^xsd:int

and hasUnit MillimetersOfMercury

- Situation would include date-time and geographical location in its definition.

- I haven't tried representing the other information yet.

3. 'Blood Pressure value recorded in a chart/EMR' is what is being called as a sign-recording? see the Arcus senilis example from Chime in the email below.

- can this instead be represented as contents of a Physical Exam record?

Physical Exam Record :=

Documentation Artifact that   outputOf   Physical Exam Procedure

One other thing that I find very peculiar is:

:mmHg

muo:measuresQuality :DiastolicBloodPressure, :SystolicBloodPressure

:DiastolicBloodPressure a snap:Quality

:SystolicBloodPressure a snap:Quality

- this seems to be backwards. Shouldn't this be inferred from the straighter assertion BloodPressure value uses mmHg as its units? 

Sivaram

[Chimezie Ogbuji]

Hey Bob.

On Fri, Jun 11, 2010 at 9:58 AM, Bob Powers <bobpo...@gmail.com> wrote:
..snip..

> The ontology-epistemology divide seems to me to be different from the
> distinction between an entity and the record of it.

Yes, they are different distinctions but - in my opinion - the two
most egregious conflations of meaning made by people who use clinical
terminology (at least that is what I have found in trying to elicit
the meaning of these terms). The representational-artifact hierarchy
is meant to address the entity and record of it distinction. The
representationOf relationship is motivated by the relationship between
a URI and what it 'identifies' as well as the relationship between a
token and what it denotes (in model theory, semiotics, etc.). It
essentially is meant to have the same semantics as the IAO's 'is
about' relationship.

> (Chime, thanks for the
> reference to this http://ontology.buffalo.edu/medo/Onto_Epist.pdf.)

No problem.

> Just
> because an entity has some of the properties of a record does not make it
> so.  Since I'm of the opinion that signs/symptoms are not records, making a
> clear distinction between sign and sign-record is very reassuring.
> The question is, how should 'sign' be constructed?  I would hope that there
> would be some formulation for what a (universal) sign is logically prior to
> its being referred to by a (universal) record.

I think it is clearer in the case of a sign than with a symptom (since
it is meant to be objective and empirical). The wiki says a sign is

'an indication of some medical fact or characteristic that may be
detected by a physician during a physical examination of a patient'.

Stedman's dictionary says it is

'Any abnormality indicative of disease, discoverable on examination of
the patient; an objective indication of disease, in contrast to a
symptom, which is a subjective indication of disease.'

So whether or not they are the manifestation of a disease, at the very
least they are observable and so are 'real things' (universals or
natural kinds). If you go with the definition that includes the
criteria that they are indicative of a disease, then it is not clear
if the indication is hypothesized, if it is a necessary component of
being a sign, or if the relationship is causal.

I tend to defer to Caroline Whitbecks assessment [1] of this topic
since it is one of the more cited works about the 'philosophy of
(clinical) medicine' and even by her account it is not entirely clear
if you can say that signs have a causal relationship with diseases:

"a contributing condition or cause differs from a necessary condition
in that its absence
does not guarantee the non-occurrence of the phenomenon in question
[...] a common error, namely that of speaking as though a proper part
of the disease process as an effect of the disease. However the
(clinical and pathological) manifestations of a disease (as opposed,
say, to the signs that one has had a disease) are not external to the
disease process. They are states of affairs which occur within the
disease process. Given the analysis of causal relations presented in
section three above, it is clear that a whole cannot be said to cause
its parts."

and..

"The rejection of the view that diseases cause the signs and symptoms
which are their characteristic manifestations does not decide the
question of whether (the presence of) a disease as a whole explains
(the presence of) its characteristic signs and symptoms"

etc. She goes into great detail to at least demonstrate that the
relationship between a disease and its manifestations (it's signs and
symptoms) is problematic. So, it seems to me, we are not left with
much to properly 'construct' the criteria of the 'sign' universal. My
question is if this is necessary for the usefulness of an ontology for
clinical medicine? The distinction doesn't seem to have much of a
value to answer informatics questions. Which is why I think providing
a way to construct the representational artifacts that represent signs
and not the things their referents is the more pragmatic approach.

> ..snip ..

> 'clinical examination'
>   realizes 'indication role'
>     'is role of' GDC
>       inheresIn 'bodily feature'
>   realizes 'clinician role'
>     'is role of' person

> ..snip ..

> This would seem to restrict GDC's to depending on ICs.  Too bad, I think
> that we really need signs/symptoms that are able to be (or are able to be
> closely tied to) occurrents so as not prematurely to model too closely.

I think I follow the challenges you outline, but I have two additional
questions: 1) what is the nature of the 'indication role' and 2) what
about signs that have not yet been observed during a clinical exam?

> So we are trying to cross the ontology-epistemology <chasm>, but with purely
> BFO-realist-ontology tools that cannot step outside the domain.

I agree.

> This is
> where, I take it, Michel's SIO solution does indeed step outside.
> Dare I suggest an 'epistemically dependent continuant'?  One that inheres in
> an entity and that has an interpretive role that is realized in an
> interpretive process.

So, the GDC would inhere in the sign that plays an 'interpretive role'
in an interpretive process? I wonder whether this is similar in some
way to the SNOMED-CT interpets relationship (which is asserted about
clinical findings):

"This attribute refers to the entity being evaluated or interpreted,
when an evaluation, interpretation or “judgment” is intrinsic to the
meaning of a concept."

>  The chasm is more fundamental than a distinction
> between continuants and occurrents, or between specific and generic
> dependency.  But maybe one can just label the thing and move on.

+1 (I'm in favor of this)

> Have no
> idea whether this is worthy or original, and it's way beyond the purview of
> the CPR itself.

agreed

>But at least it's got a target painted on its back:)
> My strong prejudice is that, whatever solution emerges, it is worth getting
> right at some granularity for the sake of the whole stack.

[1] Whitbeck, C., Causation in Medicine: The Disease Entity Model.
Philosophy of Science, Vol. 44, No. 4, pp. 619-637. The University of
Chicago Press. Dec., 1977

[Bob Powers]

Hi Chime,

I'm not pushing any view in particular here, especially for CPR, just seeing where things go.  Just a few comments inline...

On Fri, Jun 11, 2010 at 3:48 PM, Chimezie Ogbuji <chim...@gmail.com> wrote:

Hey Bob.

On Fri, Jun 11, 2010 at 9:58 AM, Bob Powers <bobpo...@gmail.com> wrote:
..snip..

> The ontology-epistemology divide seems to me to be different from the
> distinction between an entity and the record of it.

Yes, they are different distinctions but - in my opinion - the two
most egregious conflations of meaning made by people who use clinical
terminology (at least that is what I have found in trying to elicit
the meaning of these terms).  The representational-artifact hierarchy
is meant to address the entity and record of it distinction.  The
representationOf relationship is motivated by the relationship between
a URI and what it 'identifies' as well as the relationship between a
token and what it denotes (in model theory, semiotics, etc.).  It
essentially is meant to have the same semantics as the IAO's 'is
about' relationship.

Of course, representationOf is a necessary relation.  It's not in today's RO or (I haven't checked) IAO, but does CPR have to mint it?  Maybe so.  I keep expecting BFO 2.0.

> (Chime, thanks for the
> reference to this http://ontology.buffalo.edu/medo/Onto_Epist.pdf.)

No problem.

> Just
> because an entity has some of the properties of a record does not make it
> so.  Since I'm of the opinion that signs/symptoms are not records, making a
> clear distinction between sign and sign-record is very reassuring.
> The question is, how should 'sign' be constructed?  I would hope that there
> would be some formulation for what a (universal) sign is logically prior to
> its being referred to by a (universal) record.

Just so it's not a record. 

 

 

I think it is clearer in the case of a sign than with a symptom (since
it is meant to be objective and empirical).

The distinction between sign and symptom seems less important to me than establishing what they have in common.  There have been recent discussions in OGMS about someone's reporting that they were knocked out for a while.  So who's the observer?  The answer may not really matter and dwelling there might tempt one to over-modelling.

I am not surprised here.  So it sounds like we have three broad formulations to get to sign/symptom: through disease, by crossing the epistemic divide, and by reference from a record.

 

 So, it seems to me, we are not left with
much to properly 'construct' the criteria of the 'sign' universal.  My
question is if this is necessary for the usefulness of an ontology for
clinical medicine? The distinction doesn't seem to have much of a
value to answer informatics questions.  Which is why I think providing
a way to construct the representational artifacts that represent signs
and not the things their referents is the more pragmatic approach.

Is good!  My only concern here is a vague one and I don't have an example.  My question for the approach from records is: Even if one never constructs the referent to the record, would it be possible to do so?  In the future, in a different use case from what now is contemplated, after someone runs an inference engine, after the understandings of medicine change, and after what we label a sign/symptom might be newly labelled as disease/disorder itself, even after these things, can we get to a real 'bodily feature'?  Is the referent of a record a more brittle thing than a broadly modelled entity restricted under 'bodily feature'?  My writing has gone all woolly here.

> ..snip ..

> 'clinical examination'
>   realizes 'indication role'
>     'is role of' GDC
>       inheresIn 'bodily feature'
>   realizes 'clinician role'
>     'is role of' person

> ..snip ..

> This would seem to restrict GDC's to depending on ICs.  Too bad, I think
> that we really need signs/symptoms that are able to be (or are able to be
> closely tied to) occurrents so as not prematurely to model too closely.

I think I follow the challenges you outline, but I have two additional
questions: 1) what is the nature of the 'indication role' and 2) what
about signs that have not yet been observed during a clinical exam?

Don't ask took closely :)  I'm just trying to follow Michel's pattern here.  He might check whether we're allowed to have such roles.

> So we are trying to cross the ontology-epistemology <chasm>, but with purely
> BFO-realist-ontology tools that cannot step outside the domain.

I agree.

> This is
> where, I take it, Michel's SIO solution does indeed step outside.
> Dare I suggest an 'epistemically dependent continuant'?  One that inheres in
> an entity and that has an interpretive role that is realized in an
> interpretive process.

So, the GDC would inhere in the sign that plays an 'interpretive role'
in an interpretive process? I wonder whether this is similar in some
way to the SNOMED-CT interpets relationship (which is asserted about
clinical findings):

"This attribute refers to the entity being evaluated or interpreted,
when an evaluation, interpretation or “judgment” is intrinsic to the
meaning of a concept."

>  The chasm is more fundamental than a distinction
> between continuants and occurrents, or between specific and generic
> dependency.  But maybe one can just label the thing and move on.

+1 (I'm in favor of this)

> Have no
> idea whether this is worthy or original, and it's way beyond the purview of
> the CPR itself.

agreed

Agreed.