=Paper=
{{Paper
|id=None
|storemode=property
|title=Revisiting ontological foundations of the OpenEHR Entry Model
|pdfUrl=https://ceur-ws.org/Vol-897/session4-paper23.pdf
|volume=Vol-897
|dblpUrl=https://dblp.org/rec/conf/icbo/AndradeAS12
}}
==Revisiting ontological foundations of the OpenEHR Entry Model==
https://ceur-ws.org/Vol-897/session4-paper23.pdf
Revisiting ontological foundations of the OpenEHR Entry Model
André Q Andrade1,2*, Maurício B Almeida3 and Stefan Schulz2,4
1
Information Science Graduate Program, School of Information Science,
Federal University of Minas Gerais, Belo Horizonte, Brazil
2
Institute for Medical Informatics, Statistics and Documentation, Medical University of Graz, Austria
3
Theory and Management Department, School of Information Science,
Federal University of Minas Gerais, Belo Horizonte, Brazil
4
Institute of Medical Biometry and Medical Informatics,
University Medical Center, Freiburg, Germany
ABSTRACT (1) a generic information model, the reference model with
Both information models and ontologies promise ad- domain-invariant classes to be instantiated by
vantages in promoting interoperability. Recent research has (2) specific clinical models, which support semantic in-
shown the benefits of rigorous modelling in assuring large- teroperability, which are called archetypes, containing
scale consistency. specific clinical information (Beale & Heard, 2007).
In previous work we have demonstrated the feasibility Ontologies, based on the study of reality, are an alterna-
of remodelling the OpenEHR information ontology using tive solution to interoperability issues. With communication
realist ontologies, such as IAO. We here present an OWL standards they share the goal of unifying meaning across
version of the care-entry model, showing that many terms different communities, maintaining common (machine-
contained in clinical archetypes refer to reality rather than to processable) interpretations. The difference is that ontology-
information. based models are based on formal logic and are, to different
Even though not covering the domain of the infor- degrees, influenced by philosophical methodologies.
mation model (which deals with record structure, data types, The practical and pragmatic orientation of the
etc.) we have shown that the harmonization of the OpenEHR standard1, which has been described as grounded
OpenEHR standard with realist ontologies is feasible. While in an ontology of healthcare information (Beale & Heard,
useful to resolve ambiguities contained in archetype metada- 2007), closely follows medical documentation routines. In
ta definition, the proposed merged ontology also revealed contrast, ontologies developed according to realism-based
several modelling inconsistencies on published archetypes. methodologies constrain the use of some common terms in
We have also demonstrated that ambiguity in relations and clinical practice in favour of a scientific orientation (Schulz
ontological commitment can be improved by rigorous onto- et al., 2009). While realism-based ontologies were chal-
logical definitions. lenged for not being able to record all kinds of clinical data
(Dumontier & Hoehndorf, 2010; Merrill, 2010), the
1 INTRODUCTION OpenEHR entry model was found to lack the ontological
The lack of interoperability is recognized in medical in- soundness required for interoperability (Smith & Ceusters,
formatics communities as one of the main obstacles for the 2010).
full use of healthcare information systems. This issue has In previous work (Andrade & Almeida, 2011) we have
led to the creation of standards development organizations argued that the basic ontological foundation of OpenEHR
with the purpose to build consensus by proposing common- archetypes could be better represented by realist ontologies,
ly agreed message types, terms and architectural patterns. such as the Information Artifact Ontology (IAO, 2011) and
Within the realm of standards, models underlying initiatives OGMS, the Ontology for General Medical Science
like Health Level Seven International (HL7) and Open Elec- (Scheuermann et al., 2009), both based on the BFO, the
tronic Health Records (OpenEHR) try to ensure interopera- Basic Formal Ontology (Grenon et al., 2004). We now pre-
bility by defining basic templates to represent information in sent an extension of this work, in order to demonstrate the
health records. Those templates consist of a set of common feasibility of representing the OpenEHR care entry infor-
information and clinical variables that faithfully represent mation using a formal language within the framework of
health record information. The OpenEHR standard, e.g., realist ontology. We then discuss practical and modelling
defines advantages of this approach.
* To whom correspondence should be addressed: andradeaq@ufmg.br 1
http://www.openehr.org/home.html
1
Andrade, Almeida and Schulz.
This paper is structured as follows: in a brief introduc- (MacLeod, 2005). Taken as a methodology, ontological
tion we describe the OpenEHR two-level modelling ap- realism is widely used in biomedicine (Grenon et al., 2004),
proach for clinical data representation; we then compare it and its generic tenets are the following: i) there is a real
with the methodology of ontological realism. Finally, we world; ii) the reality in which we live in is part of this
propose an ontological representation of the entry types, and world; iii) we are capable of knowing the world and reality,
show practical results obtained and discuss findings. even if in an approximate way (Munn & Smith, 2008).
2 INFORMATION MODELS AND
ONTOLOGIES
Required for the development of any information sys-
tem, modelling is one of the most important aspects of soft-
ware engineering. However, despite the number of best
practices and research developed to this subject, there is not
yet an undisputed best way of representing a domain. In a
complex domain such as healthcare, it is not surprising that
many approaches have been used throughout the years.
The modelling and ontological foundations of
OpenEHR are a consequence of several previous efforts to
improve the structure and communication capabilities of Fig. 1. The OpenEHR Ontology of Recorded Infor-
Electronic Health Records. Probably due to such origins, an mation(Beale & Heard, 2007).
important principle of the OpenEHR architecture is the Our goal in this paper is not to analyse the whole stand-
separation between an ontology of reality and an ontology ard but only the care entry structure, due to its similarity and
of information (Beale & Heard, 2007). The ontology of overlap with ontologically described entities. For this pur-
information encompasses the information model and the poses, the realist approach brings two main advantages. The
domain content model (including the clinical archetypes). first is the clear separation between information entities and
All entities in the ontology of information are information real entities, which are related by the relation isAbout – e.g.
artefacts (terms, documents, images, hypotheses, orders, and the drawing of a horse is about a real horse, or a shadow on
so on) and not real clinical entities. The view on information a radiological image is about some anatomical structure.
artefacts as immaterial but nevertheless ontologically rele- That prevents inadvertent incorrect inferences of common
vant entities is gradually substituting the view of an ontolo- language statements, such as “patient blood pressure is an
gy-epistemology divide (Bodenreider et al., 2004) which observation, and all observations are created by healthcare
had emerged at a time when realist ontologies were ignoring professionals, therefore the patient blood pressure is created
the existence of information entities. by healthcare professional”. The second advantage concerns
The ontology of reality represents clinical and (patho-) the possibility of re-use of several previously developed
physiological processes, body parts, chemicals, procedures, ontologies adopting the principles of the OBO Foundry
etc. As OpenEHR makes no distinction between terminolo- (Smith et al., 2007). Those ontologies follow the same up-
gies, medical classifications and realist ontologies, this per-level ontology, the Basic Formal Ontology (Grenon et
category encompasses the International Classifications of al., 2004). This re-use promotes consensus and orthogonali-
Diseases (ICD), Logical Observation Identifiers Names and ty between ontologies, which increase robustness required
Codes (LOINC), as well as most parts of SNOMED CT. for large-scale systems, such as Electronic Health Records.
The information model is a rather complex and detailed
architecture of a generic EHR. It encompasses both defini- 3 ONTOLOGICAL REPRESENTATION
tion of records and documents (e.g. classes such as Folder, To demonstrate the compatibility between the
Composition, Section and Entry), and of the basic functions OpenEHR model and clinically oriented realist ontologies,
of software systems such as Data structure, Data type, Ac- we have created an OpenEHR information ontology accord-
cess, Version, etc.). The care entry model “define the se- ing to the OGMS guidelines. To properly place each class,
mantics of all the „hard‟ information in the record” (Beale & we took into consideration the natural language description
Heard, 2007), and represents information recorded during a and basic rationale used to develop the entry types. In such
medical encounter. Figure 1 shows a graphical representa- rationale, the history classes (“observation” and “action”)
tion of the ontology leading to the care entry model. represent statements about the past events of the individual
Following a quite different principle, realist ontologies subject of record. This includes the description of currently
are based on the philosophical study of reality. The term observed characteristics, based on the fact that their appear-
“realism” in Philosophy is widely used and controversial ance necessarily precedes the observation. The “evaluation”
2
Revisiting ontological foundations of the OpenEHR Entry Model
classes represent current assessment by the attending health due to its stability and simplicity. Some examples are shown
professional, including “diagnosis” and “prognosis”, as well in Table 1. It was modelled as information about blood
as the representation of the imagined future, like “goals” pressure, which is understood as the quality of a portion of
and “scenarios”. “Instructions” represent future events that blood that exerts pressure toward an artery. For didactical
should take place as prescribed by the health professional. purposes, the ontological aspects of blood pressure were
The proposed merged ontology can be seen in figures 2 simplified (Goldfain et al., 2011; Kumar & Smith, 2003).
and 3. While the rationale is quite different, the IAO proved Likewise, systolic and diastolic blood pressure measure-
capable of faithfully representing the meaning of each in- ments were easily modelled and their information status
formation type. Observation is a Data item resulting from allows unambiguous assignment of data values and types.
the medical encounter, being a description of an entity, Some challenges arose while modelling reference to the
usually, the patient. By classifying the other classes accord- measurement procedures, rather than to the pressure itself
ing to their intended outcome, we merged the Proposal (e.g. the representation of the position of the patient while
classes under Objective specification and Instruction classes being measured - sitting, upright - and the size of the cuff
under Plan specification. Finally, Action was represented as used for measurement). This required explicit representation
a special type of report, since it necessarily describes a pro- of the measurement as a process having the patient as partic-
cess that has the patient as participant. ipant. Also, patient positions could be adequately represent-
ed as qualities of the patient, who is a participant of the
measurement process.
However, several epistemic entities were not success-
fully modelled, as they are not properly representable in
realist ontologies. As an example, consider the metadata
“confounding factors”, defined as “Comment on and record
other incidental factors that may be contributing to the blood
pressure measurement. (…), level of anxiety or 'white coat
syndrome'; pain or fever; changes in atmospheric pressure
etc.” Events such as pain and changes in atmospheric pres-
sure have little or nothing in common that could map them
to one category in an ontology. E.g. a confounding factor
can be a process, a disposition, or a quality. Whether such
Fig. 2. The OpenEHR Care Information branch “non-ontological” classes – characterized as “defined clas-
ses” by (Smith et al., 2006) – belong in an ontology at all, is
contentious. However they can represented by logical defi-
nitions in an OWL model (Schulz et al., 2011).
Finally, some attributes that are specific to medical
practice, such as “Diastolic endpoint”, have unclear repre-
sentation in real ontologies. Defined in the Blood Pressure
Archetype as a metadata allowing the user to “record which
Korotkoff sound2 is used for determining diastolic pressure
using auscultative method”, this attribute depends on train-
ing and individual interpretation to be defined, and lack
ontological status (Andrade & Almeida, 2011). This is
clearly shown by the lack of rigor in the distinction between
Fig. 3. The OpenEHR entry directive information branch the 4th and 5th sounds, which refer to perceptive capabilities
Using this new ontology, we proceeded to analyse one ar- of the actor, defined as (our emphasis) “phase IV, sounds
chetype of each main branch (observation, action, evalua- become muffled and softer; and phase V, sounds disappear
tion and instruction) contained in the OpenEHR Clinical completely. The fifth phase is thus recorded as the last au-
Knowledge Manager (OpenEHR Foundation, 2010). Our dible sound” (Pickering et al., 2005). However, such attrib-
criteria for selection were based on published status and utes can also be represented by logical definitions, and
frequency in medical records. The findings are summarized should be subject of further investigation.
in the next section.
3.1 Observation Archetypes
We analysed the most commonly used example of an 2
Sounds heard during measurement of blood pressure.
observation archetype, viz. the Blood Pressure Archetype,
3
Andrade, Almeida and Schulz.
Class name openEHR-EHR- tional modelling of reality, e.g. roles and parts, in order to
OBSERVATION.blood_pressure.v1 express the real entities represented in the information arte-
Elucidation The local measurement of arterial blood pres- facts.
sure which is a surrogate for arterial pressure 3.2 Action Archetypes
in the systemic circulation. Most commonly, We examined the Medication Action Archetype, which
use of the term 'blood pressure' refers to represents one of the most commonly described healthcare
measurement of brachial artery pressure in the interventions. Its precise reconstruction was not straightfor-
upper arm. ward, as it included states that contradict the existence of a
Axiomati- is_aboutsome arterial_blood_pressure process, e.g.Cancelled or Postponed states. In other words, a
zation is_output_of some cancelled process is not a kind of process, since the process
blood_pressure_measurement_process never actually took place. Therefore, a different treatment is
has_part some required, as only plans about medication administration
blood_pressure_measurement_datum processes can be cancelled or postponed, not the processes
Superclass 'physical examination finding' themselves (Raufie et al., 2011; Schulz & Karlsson, 2011).
Class name Systolic_blood_pressure_data Furthermore, because this archetype includes medica-
Elucidation Represents the Systolic attribute, defined as tion-specific information, it is not clear what kind of relation
Peak systemic arterial blood pressure - meas- between the action and the medication holds. Since this
ured in systolic or contraction phase of the template has information such as Name of medication as
heart cycle. well as Reason for ceasing the medication, an explicit defi-
Axiomati- is_output_of some nition of those relations in the archetype is required before
zation Systolic_blood_pressure_measuring_process further modelling.
Superclass blood_pressure_measurement_datum
3.3 Evaluation Archetypes
Class name human_position_configuration
Elucidation Represents the Position attribute, defined as For this analysis, we used two archetypes. The first is a
The position of the subject at the time of publicly published evaluation archetype, the Clinical Synop-
measurement. sis Archetype. While extremely simple, this class proved
conformant to its information status, being defined as “nar-
Axiomati- is_quality_of some human_being
rative summary or overview about a patient, specifically
zation
from the perspective of a healthcare provider, and with or
Superclass configuration (subClass of quality)
without associated interpretations.” This definition suggests
Class name blood_pressure_measurement_process that a class such as OGMS Clinical Picture perfectly de-
Elucidation Represents the actual process of measurement scribes its meaning, though not as an overarching concept to
that will result in the blood pressure observa- diagnosis or objective specifications.
tion. It is not directly referred in the OpenEHR The second one is an archetype that demonstrates the
archetype. Extended and modified from the classification as Objective specification, the Goal Arche-
Vital Sign Ontology. type. It conformed to modelling, requiring specification of
Axiomati- has_participant some the objective intended, along with the time where it is ex-
zation (human_being pected and the standard that will evaluate its success. Over-
and (has_quality some all, this class appears to be adequately represented.
human_position_configuration)
and (has_quality some sleep_status) 3.4 Instruction Archetypes
and (has_role some patient_role) For this analysis of this archetype category we evaluat-
and (has_part some artery_wall)) ed the medication order, being one of the most common
Superclass ‘planned process’ instructions. It is defined as “an order or instruction created
Table 1: Mapping the Blood Pressure Archetype to OWL by a clinician which specifies which medication to take,
when, for how long etc.” It is directly related to the medica-
Overall, the ontological interpretation of the blood pres- tion action archetype by an item called Medication activity
sure archetype revealed definitions that could only be un- (See section 3.2). This was suitable for ontological represen-
derstood by a domain specialist. It also made clear that the tation since it can be shown that both actions and instruc-
archetype is information about the patient, the examination tions are about processes. While the actions are kinds of
procedures, the examination artefacts used in the procedure processes, the instructions are plans that specify a process
(real entities) and the health professional‟s interpretation of type which may be or not instantiated in the future. Not only
the process (information entities). As such, it requires addi- the ontological representation faithfully describes the enti-
4
Revisiting ontological foundations of the OpenEHR Entry Model
ties about medication, but also makes clearer the distinction Grenon, P., Smith, B. and Goldberg, L. (2004) Biodynamic ontology:
between a suspended plan (the medication will not be ad- applying BFO in the biomedical domain. In Pisanelli, D.M. (ed),
ministered to the patient) and a suspended administration Ontologies in Medicine. IOS Press, Amsterdam, pp. 20-38.
(the administration process started but was stopped before IAO (2011) Information Artifact Ontology.(last accessed in January 30th
completion). 2011)..
4 CONCLUSION Kumar, A. and Smith, B. (2003) The ontology of blood pressure: A case
Though not addressing those aspects of the information study in creating ontology partitions in biomedicine..
have presented examples which demonstrate that the har- MacLeod, M.C. (2005) Universals.(last accessed in June 28th
monization of the OpenEHR standard by representing arche- 2011)..
types as realist ontologies is feasible. While useful to detect Merrill, G.H. (2010) Ontological realism: Methodology or misdirection?,
and fix ambiguities in archetype metadata definitions, the Applied Ontology, 5, 79-108.
merged ontology also revealed several modelling inconsist- Munn, K. and Smith, B. (2008) Applied Ontology. An Introduction. Ontos
encies on published archetypes. We have also shown that Verlag, Frankfurt/Paris/Lancaster/New Brunswick, pp. 342.
ambiguity in relations and ontological commitment can be OpenEHR Foundation (2010) Clinical Knowledge Manager.(last accessed
improved by providing rigorous ontological definitions. in January 30th 2012)..
Future work should focus on standard ways of ontologically Pickering, T.G., Hall, J.E., Appel, L.J., Falkner, B.E., Graves, J., Hill,
representing epistemic and interpretative clinical infor- M.N., Jones, D.W., Kurtz, T., Sheps, S.G. and Roccella, E.J. (2005)
mation, together with their linkage to reality entities. Also, Recommendations for blood pressure measurement in humans and
the precise logical formulation of value constraints, data experimental animals: Part 1: blood pressure measurement in humans,
types and cardinality requires further studies, to guarantee Hypertension, 45, 142-161.
universal interpretability of these kinds of representation. Raufie, D., Schulz, S., Schober, D., Jansen, L. and Boeker, M. (2011)
Redesigning an Ontology Design Pattern for Realist Ontologies.
ACKNOWLEDGEMENTS OBML. Berlin, Germany.
Scheuermann, R.H., Ceusters, W. and Smith, B. (2009) Toward an
This work is supported by the Hemominas Foundation (Belo
Ontological Treatment of Disease and Diagnosis. 2009 AMIA Summit
Horizonte, Brazil). AQA was being financed by Coordenação de
on Translational Bioinformatics. San Francisco, CA, pp. 116-120.
Aperfeiçoamento de Pessoal de Nível Superior (Brazil) – Programa
Schulz, S., Brochhausen, M. and Hoehndorf, R. (2011) Higgs bosons, mars
de Doutorado no País com Estágio no Exterior, process number
missions, and unicorn delusions: How to deal with terms of dubious
2380-11-0, during the writing of this paper.
reference in scientific ontologies. Proceedings of the 2nd International
Conference on Biomedical Ontology. 2011. CEUR-WS, Buffalo, NY,
REFERENCES USA, pp. 183-190.
Andrade, A.Q. and Almeida, M.B. (2011) A realism-based analysis of the Schulz, S. and Karlsson, D. (2011) Records and situations. Integrating
OpenEHR entry model. Proceedings of the MIXHS'11. ACM, contextual aspects in clinical ontologies. In Shah, N., et al. (eds),
Glasgow, Scotland, UK, pp. 51-58. Proceedings of the BioOntologies SIG at ISMB/ECCB 2011. Vienna,
Andrade, A.Q. and Almeida, M.B. (2011) Realist representation of the Austria, pp. 49-52.
medical practice: an ontological and epistemological analysis. Schulz, S., Stenzhorn, H., Boeker, M. and Smith, B. (2009) Strengths and
Proceedings of the 4th Ontobras. Gramado, Brazil. limitations of formal ontologies in the biomedical domain, RECIIS Rev
Beale, T. and Heard, S. (2007) Architecture Overview. OpenEHR Electron Comun Inf Inov Saude, 3, 31-45.
Foundation, pp. 87. Smith, B., Ashburner, M., Rosse, C., Bard, J., Bug, W., Ceusters, W.,
Beale, T. and Heard, S. (2007) An ontology-based model of clinical Goldberg, L.J., Eilbeck, K., Ireland, A., Mungall, C.J., Leontis, N.,
information, Stud Health Technol Inform, 129, 760-764. Rocca-Serra, P., Ruttenberg, A., Sansone, S.A., Scheuermann, R.H.,
Bodenreider, O., Smith, B. and Burgun, A. (2004) The Ontology- Shah, N., Whetzel, P.L., Lewis, S. and Consortium, O.B.I. (2007) The
Epistemology Divide:A Case Study in Medical Terminology. In Varzi, OBO Foundry: coordinated evolution of ontologies to support
A. and Vieu, L. (eds), 3rd Conference on Formal Ontology in biomedical data integration, Nature Biotechnology, 25, 1251-1255.
Information Systems (FOIS 2004). Turin. Smith, B. and Ceusters, W. (2010) Ontological realism: A methodology for
Dumontier, M. and Hoehndorf, R. (2010) Realism for scientific ontologies. coordinated evolution of scientific ontologies, Applied Ontology, 5,
Proceedings of the Sixth International Conference on Formal 139–188.
Ontology in Information Systems (FOIS 2010). IOS Press, Toronto, Smith, B., Kusnierczyk, W., Schober, D. and Ceusters, W. (2006) Towards
Canada, pp. 387-399. a reference terminology for ontology research and development in the
Goldfain, A., Smith, B., Arabandi, S., Brochhausen, M. and Hogan, W.R. biomedical domain. KR-MED 2006. pp. 57-66.
(2011) Vital Sign Ontology. In Shah, N., et al. (eds), Proceedings of
the BioOntologies SIG at ISMB/ECCB 2011. Vienna, Austria.
5