CPR-IAO-OGMS overlaps - Githubissues

From: Chimezie Ogbuji <chimezie@gmail.com>
Date: Sunday, November 8, 2009
Subject: Some discussion regarding the overlaps between CPR / IAO /
OGMS /  BioTop
To: cpr-ontology <cpr-ontology@googlegroups.com>
Cc: arabans@gmail.com

In its current form, the CPR ontology imports and uses BioTop.  BioTop
re-uses some terms from the Information Artifact Ontology (IAO).  In
addition, the Ontology for General Medical Science (OGMS) has recently
surfaced and is based the Toward an Ontological Treatment of Disease
and Diagnosis paper that the CPR ontology is also based on.  In trying
to reconcile these overlaps, I took some notes (below) about the
discrepancies.

# Duplicated Terms #

Bodily feature is in both OGMS and in CPR.  However, in CPR the term
is properly axiomatized whereas in OGMS it is a primitive class.

Sign and vital sign are defined in both OGMS and in CPR.  In CPR,
sign's class definition reflects the human readable label given to the
same term by OGMS ("A bodily feature of a patient that is observed in
a physical examination and is deemed by the clinician to be of
clinical significance") whereas it is an undefined primitive term in
OGMS.

OGMS  has a notion of a manifestation of a disease
(ogms:DiseaseManifestation) that is a primitive class with an empty
class definition.  In CPR, on the other hand, the disease class
(cpr:pathological-disposition) has a class definition which includes a
constraint that it is the cause of either a cpr:pathological-process
or a cpr:symptom, both of which match the semantics of the OGMS notion
of a manifestation of a disease.  Likewise, the OGMS disease class is
a primitive with an empty class definition, whereas the CPR term has
(in addition to its manifestations / side effects as mentioned above)
constraints regarding: morphology, etiological characteristics, and
anatomic location.

The pre-clinical and clinical suffixes used with the subclasses of
ogms:DiseaseManifestation are epistemological distinctions that at
best should not be included in an ontology for medical record concepts
[Olivier et.al.] and at worst is covered by where they fall on the
timeline.  Examples of situations involving pre-clinical and clinical
disease manifestations would help elucidate the need (or lack thereof)
to adopt these terms from OGMS.

CPR has a notion of a clinical-phenotype which is a special kind of
aggregate-bodily-feature.  It inherits the constraint (from its
parent) that it is composed of cpr:bodily-features (via has_part).
OGMS has a more general Phenotype term which has an empty class
description despite the fact that the human readable definition
includes the constraints that are defined on its CPR counterpart.
Examples of clinical situations involving phenotypes would help in
identifying the distinctions that are relevant to the domain and can
be incorporated into the class definitions of these terms.

# Significant Semantic Discrepancies #

The Information Artifact Ontology (IAO) has an information content
entity class.  It is a generically dependent continuant which suggests
that there is some independent continuant that bears it.  BioTop has
an information entity class which seems to fit this role.  In BioTop,
information entities have neither mass nor volume and do not inhere in
entities.   Currently, the key CPR representational-artifact class is
defined a special kind of information entity.  It is an open question
as to whether this is the appropriate ontological commitment for this
term.

cpr:representational-artifact is defined as a "representation that is
fixed in some medium in such a way that it can serve to make the
cognitive representations existing in the minds of separate subjects
publicly accessible in some enduring fashion. Examples are: a text, a
diagram, a map legend, a list, a clinical record, or a controlled
vocabulary."

This definition from [Smith et.al.] was meant to address the
distinction between first acts, secondary acts, and the resulting
clinical artifacts that persist in a medium (an EHR).  This
distinction is problematic in information models such as the HL7 RIM.
For example, a surgical procedure would be the first act (a clinical-
act), the recording of this procedure by the surgeon (or a subordinate
clinician) would be the second act (can also be referred to as an
informatics process - credit Helen Chen), and the entry in the
patients EHR would be the representational-artifact.  Finally, this
term was also meant to inherit the semiotic (semiotics: the study of
signs and symbols and their use or interpretation.) approach of
separating symbols from the entities they denote or represent.  A
cpr:representationOf predicate is used to capture this.  It is
questionable whether introducing a predicate that captures the
semantic relationship between a symbol and what it denotes is
appropriate if the underlying knowledge representation has a formal
model theoretic semantics (which already provides a mechanism for
interpreting statements made in this language by relating the terms in
the statements to things in the world).

In any case, the entry in the patients EHR can be said to 'represent'
the surgical operation.  Provenance information about the secondary
act (for example, who recorded it, when they recorded it, and the
source of that recording) can be captured using a vocabulary such as
Dublin Core.

In this approach, the representational artifact is a physical thing
since (as the definition suggests) it is 'fixed' in a medium and
serves as a proxy for cognitive representations that are not physical
and can be thought of as being equivalent to the IAO's information
content entities.  However, in a realist approach to formal clinical
informatics, cognitive representations are not relevant

The class definition for representational-artifact does not capture
the constraint that is fixed in a medium.  This could best be
represented by making it a generically dependent continuant where its
bearer is the digital media it is recorded or fixed in.  This is
currently not the case.  Note this dependence is different from that
between IAO's information content entity and its bearer an information
entity.

OGMS has a Representation class (a generically dependent continuant)
that does not have a definition and is the ancestor of clinical
finding, clinical history, image finding, diagnosis, laboratory
finding, pre clinical finding, and clinical picture - all of which
have a corresponding term in CPR each of which include a class
definition.
Original issue reported on code.google.com by albertgo...@gmail.com on 8 Nov 2009 at 4:15
nvaldivi / ogms

CPR-IAO-OGMS overlaps #36
