<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Archiving and Interchange DTD v1.0 20120330//EN" "JATS-archivearticle1.dtd">
<article xmlns:xlink="http://www.w3.org/1999/xlink">
  <front>
    <journal-meta />
    <article-meta>
      <title-group>
        <article-title>The CEN ISO Standard Categorial Structure as a Top-Level Set of Constraints for Ontology Disambiguation</article-title>
      </title-group>
      <contrib-group>
        <contrib contrib-type="author">
          <string-name>Jean-Marie Rodrigues</string-name>
          <email>rodrigues@univ-st-etienne.fr</email>
          <xref ref-type="aff" rid="aff0">0</xref>
          <xref ref-type="aff" rid="aff1">1</xref>
          <xref ref-type="aff" rid="aff4">4</xref>
        </contrib>
        <contrib contrib-type="author">
          <string-name>Stefan Schulz</string-name>
          <xref ref-type="aff" rid="aff2">2</xref>
          <xref ref-type="aff" rid="aff3">3</xref>
        </contrib>
        <contrib contrib-type="author">
          <string-name>Cédric Bousquet</string-name>
          <xref ref-type="aff" rid="aff0">0</xref>
          <xref ref-type="aff" rid="aff1">1</xref>
        </contrib>
        <contrib contrib-type="author">
          <string-name>Julien Souvignet</string-name>
          <xref ref-type="aff" rid="aff0">0</xref>
          <xref ref-type="aff" rid="aff1">1</xref>
        </contrib>
        <aff id="aff0">
          <label>0</label>
          <institution>Department of Public Health CHU University of Saint Etienne</institution>
          ,
          <country country="FR">France</country>
        </aff>
        <aff id="aff1">
          <label>1</label>
          <institution>INSERM U 872</institution>
          ,
          <addr-line>Eq. 20, Paris</addr-line>
          ,
          <country country="FR">France</country>
        </aff>
        <aff id="aff2">
          <label>2</label>
          <institution>Institute for Medical Biometry and Medical Informatics, University Medical Center</institution>
          ,
          <addr-line>Freiburg</addr-line>
          ,
          <country country="DE">Germany</country>
        </aff>
        <aff id="aff3">
          <label>3</label>
          <institution>Institute for Medical Informatics</institution>
          ,
          <addr-line>Statistics, and Documentation</addr-line>
          ,
          <institution>Medical University</institution>
          ,
          <addr-line>Graz</addr-line>
          ,
          <country country="AT">Austria</country>
        </aff>
        <aff id="aff4">
          <label>4</label>
          <institution>WHO-FIC Collaborative Centre for International Classifications in French Language</institution>
          ,
          <addr-line>Paris</addr-line>
          ,
          <country country="FR">France</country>
        </aff>
      </contrib-group>
      <abstract>
        <p>Since 1990 CEN TC 251 and ISO TC 215 have developed an approach named Categorial Structure. It is a logic-based language that aims at standardising the upper level structure of a terminological model rather than agreeing on a reference clinical terminology or on a language-independent biomedical ontology. Since 2000 methods, tools and techniques based on upper level ontology and description logic formalism have been developed in the Semantic Web and the bio-ontology communities. The objective of this paper is to analyse the relation between the two approaches in order to promote the complementary use of ontology and structured information model tools in the harmonisation between biomedical terminologies and to improve semantic interoperability.</p>
      </abstract>
    </article-meta>
  </front>
  <body>
    <sec id="sec-1">
      <title>INTRODUCTION</title>
      <p>
        Standardisation in Health Informatics started in the U.S.
with the HL7 user group. The European Standard Body
CEN TC 251 WG2 (Comité Européen de Normalisation
Technical Committee 251 Working Group 2) and later the
International Organisation for Standardization (ISO) TC 215
WG3 elaborated and developed a standard approach for
biomedical terminology named Categorial Structure
(CAST)
        <xref ref-type="bibr" rid="ref1">(Rodrigues, 2008)</xref>
        , mainly based on the
ontologydriven tools developed in the nineties as a result of the
GALEN project
        <xref ref-type="bibr" rid="ref2">(Rector,1995)</xref>
        .
      </p>
      <p>
        Since 2000, ontology-driven methods, tools and techniques
have been developed in the Semantic Web and the
bioontology communities. Numerous research groups have
moved towards upper-level ontologies (ULO) in order to
manage the organization, integration and interoperability of
biomedical information in a more principled way. Not only
in research circles, but also in the medical terminology
business, there is an increasing belief that using formal
ontology approaches helps improve the quality of
terminological systems. Description logics
        <xref ref-type="bibr" rid="ref3">(Baader, 2007)</xref>
        often
using Web Ontology Language (OWL)
        <xref ref-type="bibr" rid="ref4">(Horridge, 2009)</xref>
        has become a quasi-standard for formal ontologies, which
intend to describe (as much as possible) the consensus on
the nature of entities in a given scientific domain,
independently of linguistic or conceptual variation.
      </p>
      <p>
        The objective of this paper is to analyse the relation between
the two approaches, viz. upper-level ontologies (ULO) vs.
Categorial Structure (CAST), to ease the use of applied
ontology to harmonise different biomedical terminologies.
In a first part we present the definition and specifications of
the CAST standard methodology
        <xref ref-type="bibr" rid="ref1">(Rodrigues, 2008)</xref>
        ;
whereas a second part is dedicated to upper level ontologies
and the alignment of the CAST of Patient Safety (PS)
Conceptual Concept Framework with the top-level ontology
BFO
        <xref ref-type="bibr" rid="ref5">(Smith, 2011)</xref>
        and the top-domain ontology
BioTop (
        <xref ref-type="bibr" rid="ref6">Beißwanger, 2008</xref>
        ). In a third part we present the
CAST application to the ICD 11 revision. Finally we will
discuss the relation between the CAST and another ontology
approach to disambiguate healthcare terminologies.
2
      </p>
    </sec>
    <sec id="sec-2">
      <title>CEN ISO CATEGORIAL STRUCTURE</title>
    </sec>
    <sec id="sec-3">
      <title>APPROACH</title>
      <p>
        The CEN Categorial Structure was defined, as presented
in
        <xref ref-type="bibr" rid="ref1">(Rodrigues, 2008)</xref>
        , as a minimal set of constraints to
represent a biomedical terminology in a given health care
domain with the goal to safely communicate. It is a
definition of a minimal semantic structure or ontology framework
describing the main properties of the different artefacts used
as terminology (controlled vocabularies, nomenclatures,
reference terminologies, coding systems and
classifications): a model of knowledge restricted to 1) a goal, 2) a list
of semantic categories, 3) the list of semantic links or
relations between semantic categories constrained by their
associated semantic categories, and 4) the minimal constraints
allowing the generation and the validation of well-formed
terminological expressions. As a consequence any
biomedical terminology artefact claiming conformance to this
standard shall attach with the data sent the Categorial
Structure of the terminology used. The Categorial Structure shall
satisfy the four constraints, but it can add more constraints.
For instance, the CAST for terminological systems of
surgical procedures
        <xref ref-type="bibr" rid="ref7">(Rodrigues, 2011)</xref>
        specifies that:
(1) The goal is to model surgical procedures,
(2) The main semantic categories are Human Anatomy,
      </p>
      <p>Deed, Interventional Equipment and Lesion;
(3) The semantic links are has_object,
has_sub_surgicaldeed, has_means...
has_site,
(3.1) has_object is authorised between Deed and
Human anatomy or Interventional Equipment or Lesion;
(3.2.) has_site is authorised between Interventional
equipment or Lesion and Human anatomy;</p>
      <p>(3.3) has_means is authorised between Deed and
Human anatomy, Interventional equipment or Lesion;
(3.4) has_sub_surgicaldeed
Deed and Deed;
(4) The minimal constraints required are:
is authorised between
(4.1) A Deed and has_object shall be present;
(4.2) Human anatomy shall always be present either
with the relation has_object or with a has_site;
(4.3) Use of Lesion shall be restricted to macroscopic
lesion and to cases where it allows differentiating the
procedure from procedures using the same deed and the
same human anatomy;
(4.4) When has_sub_surgicaldeed is used, the Deed on
the right side of the semantic link must be conform to
the rules 3.1, 3.2 and 3.3.</p>
    </sec>
    <sec id="sec-4">
      <title>UPPER LEVEL ONTOLOGY APPROACH 3</title>
      <p>3.1</p>
      <p>
        BFO
Basic Formal Ontology (BFO) is a philosophically inspired
top-level ontological framework
        <xref ref-type="bibr" rid="ref5">(Smith, 2011)</xref>
        , which
provides general categories as a basis for domain ontologies of
a given level of granularity.
      </p>
      <p>
        BFO is divided into two main parts: Continuants (entities
which continue to exist through time), and Occurrents
(entities which exist in full in a single instant of time). It
provides a coherent and unified understanding of basic
ontological entities that are fundamental to describe a science-based
reality, and it allows integrating domain terminologies
especially biomedical ones. BFO v1.1 includes 39 classes. This
version does not include any relations, an ontology of
relationships such as OBO Relation Ontology (RO)
        <xref ref-type="bibr" rid="ref8">(Smith
2005)</xref>
        . A new release of BFO (v2) is currently under
development and should incorporate RO.
3.2
      </p>
      <sec id="sec-4-1">
        <title>BioTop</title>
        <p>
          BioTop is a top-domain ontology for biomedicine,
represented in OWL-DL. It provides a class and a relation
hierarchy, together with description logics axioms
          <xref ref-type="bibr" rid="ref3">(Baader, 2007)</xref>
          .
Furthermore, BioTop can serve as a top-level model for
creating new ontologies for more specific domains or as aid
for aligning or improving existing ones. BioTopLite is a
lighter experimental version. BioTopLite has 49 classes and
50 object properties (including inverse properties) and
presents only top level classes with a low granularity level.1
BioTop is inspired both by BFO
          <xref ref-type="bibr" rid="ref5">(Smith, 2011)</xref>
          and DOLCE
          <xref ref-type="bibr" rid="ref9">(Gangemi, 2002)</xref>
          , providing bridging ontologies to both.
3.3
        </p>
      </sec>
      <sec id="sec-4-2">
        <title>BFO and BioTop vs. Categorial Structure</title>
        <p>Analyzing the Categorial Structure in the light of
description-logics based upper ontologies we find so many
parallelisms that the hypothesis is allowed that the Categorial
Structure can be interpreted, by and large, in terms of description
logics axioms.</p>
        <p>For instance, the basic categories can be declared as disjoint
classes, e.g. Human Anatomy and Deed subclassOf Nothing
(cf. (1) in section 2).</p>
        <p>Semantic links in the model can be considered object
properties with domain and range restrictions. E.g., has_means
can be considered an OWL object property with its
respective domain and range restrictions, according to (3.3):
Thing subclassOf inverse (has_means) only Deed
Thing subclassOf has_means only
(Human anatomy or InterventionalEquipment or Lesion)
Minimal constraints in the Categorial Structure model
correspond to existential axioms, e.g. (3.1):</p>
        <p>Deed subclassOf has_object some Thing
or more complex (3.2):</p>
        <p>
          Deed subclassOf ((has_object some HumanAnatomy) or
(has_site some HumanAnatomy))
One major difference between the Categorial Structure and
description logics is that OWL object properties are
exclusively relations between individuals (hence they require
quantifiers whenever used in axioms including classes),
whereas the links in the Categorial Structure hold between
classes. However, the links between classes can be defined
by means of the links between individuals in a similar way
as done in the context of RO
          <xref ref-type="bibr" rid="ref8">(Smith 2005)</xref>
          2:
        </p>
        <p>A rel B =def A subclassOf rel some B
3.4</p>
      </sec>
      <sec id="sec-4-3">
        <title>PS-CAST Mapping to ULO</title>
        <p>
          Following the WHO Department of Patient Safety report on
a conceptual basis (a list of terms and definitions of patient
safety concepts3
          <xref ref-type="bibr" rid="ref10">(Runciman, 2009)</xref>
          named International
Classification for Patient Safety (ICPS), an ontological
representation using the Categorial Structure method,
named PS-CAST
          <xref ref-type="bibr" rid="ref11">(Souvignet, 2011)</xref>
          was issued. For
example, here a representation of the Fall incident :
has mereologic relation
has cause: (Agent: Physical Environment or
Health Services: health Intervention)
has circumstance: (Circumstance: Contributing
Factors: Staff factors or Mitigating Factors:
Effective protocol available)
1 Both versions of January 29, 2012, available at http://purl.org/biotop
2 Relation between classes/concepts: italics, between individuals: bold
3 http://www.who.int/patientsafety/taxonomy/en/
has consequence: (Harm: Patient Outcomes:
Fracture or Organizational Harm: Legal
Ramifications)
has characteristics
has type incident: (Incident Type: Fall)
has defining characteristics: (Incident
Characteristics: sudden, unintended, uncontrolled downward
displacement of a patient body to the ground)
has associative relation
has detection: (Detection)
has location: (Care_Setting: Hospital)
has people involved: (Person: Health Care
Profesional)
PS-CAST aims to integrate more granular value sets for
subsequent development and to test the robustness of the
CAST across different PS recording systems in the world it
has been decided to try to map it to upper level ontologies.
Two manual alignments were performed; first there was
between classes of PS-CAST with BFO classes for BFO is
the most generic ULO. The second step was to align both
PS-CAST classes and relations with BioTop which is a
domain ontology.
        </p>
        <p>The alignment method, however, remained the same for
both tasks. First, we performed an analysis of the class
definitions proposed in the ontologies (BFO and BioTop),
where relevant classes were kept. Then a search for
connections between these classes and PS-CAST was performed.
Each alignment proposed was then reviewed by a Patient
Safety expert. If class definitions were unclear or inaccurate,
or if a relation was questionable, new proposals were made.
The mapping was based on the OWL-DL versions of both
ontologies. It used taxonomic subsumption (“is-a”) (A
subclassOf B), and equivalence (A equivalentTo B).</p>
        <p>The mappings were considered complete after a few
iterations when no more changes were needed. We have ensured
that no violation of domains and ranges were made.
Table 1 is an excerpt of the two proposed mappings for
PSCAST with BFO and with BioTop.</p>
        <p>PS-CAST BFO (link type) BioTop (link type)
Incident Process (subclass) Process (subclass)
Incident_Type Disposition (subclass) Disposition (subclass)
Circumstance Process OR Independent Condition (subclass)</p>
        <p>
          Continuant (subclass)
lssseaCPCDeaertrseeoc_tnSioentting SPRirotoelec(e(sseusqbu(csilvuaasbslce)lnatssc)lass) ARMcoatliteoer(neiaq(sluuOibvbcajlleaecsntst)(csluabscs)lass)
Harm Process (subclass) Condition (subclass)
Anatomy Independant Continuant (StructuredBiologicalEntity OR
(subclass) ImmaterialObject) (subclass)
… … …
has_Cause - causedBy (equivalent relation)
itseerropPhhhhaaaassss____ILCMnooeccniaadsnteesinqotun_eTnycpee ---- (chphsaraauuossbLRscreeeoesslcaas(uluteiiszaqoal(unletiy)iqvo-auRnlieve(nlasatluetrbenerdtleaTrltaeiotolianot)ino)n)
… - …
As a general observation, the naming of several PS-CAST
classes suggests that these classes were defined rather by
functional than by ontological criteria. For instance, a
Circumstance is difficult to describe. E.g., a certain place can
be a circumstance of an adverse event just as an event or a
disposition. This ambiguity could be expressed by mapping
“PSCAST:Circumstance” to a disjunctive expression like
“bfo:Process or bfo:IndependentContinuant or
“bfo:Disposition”. In BioTop, such a disjunctive class is
already there, due to the need to represent ambiguous
clinical terms like “allergy” or “tumor”, for which different
readings (processual, material, dispositional) coexist
          <xref ref-type="bibr" rid="ref12">(Schulz, 2011)</xref>
          . A difficult distinction is also the one
between Incident and Incident_Type due as well to the
ontology ambiguity of incident which has to be considered as a
disjunctive expression. The ontological distinction between
Incident and Incident_Type proved difficult. Our current
hypothesis is that the intended meaning of "Incident_Type"
is a dispositional one, i.e. an Incident_Type is a realizable
entity which can be realized by a corresponding Incident.
On one hand, due to its biomedical domain coverage,
BioTop provides finer alignment of some concepts such as
Anatomy, which is aligned in BFO with “BFO:Independent
Continuant” and in BioTop Lite with (“BioTop:Material Object
&gt; BioTop:Polymolecular composite entity &gt; BioTop:structured
biological entity” OR “BioTop:Immaterial Object”). On the
other hand, for PS-CAST domain independent classes,
BioTop is less suitable. Some classes do not seem to find a
place in the ontological tree and/or did not match exactly the
definitions while BFO was more accurate, for example,
“PSCAST:Circumstance”, mapped with “BFO:Process context”.
The mapping of the relations was done only for Bio Top. It
also posed difficulties such as the relation
“PSCAST:has_means”, which is part of a process, but there
is no similar relation in Bio Top, the link was made on a
high-class level “BioTop:procesually related to”.
We have not yet compared the BioTop mapping with both
BFO and RO to verify our alignments and to adjust them.
4
        </p>
      </sec>
    </sec>
    <sec id="sec-5">
      <title>CAST APROACH TO ICD-11 REVISION</title>
      <p>
        The World Health Organization (WHO) has initiated the
revision process of the International Classification of
Diseases4 (ICD-11) in 2007. Different from past revisions done
by WHO-FIC collaborating centres, the ICD-11 authoring
process, which involves a large community of clinical
experts, is supported by ontology-driven tools
        <xref ref-type="bibr" rid="ref14">(Tudorache,
2010)</xref>
        <xref ref-type="bibr" rid="ref15">(Rodrigues, 2010)</xref>
        . Another difference is the
distinction between a multi-hierarchical ICD foundation
component (FC) as a blend of ontology and information model to
be used as a basis for target specific linearization products.
4 www.who.int/classifications/icd/
4.1
      </p>
      <sec id="sec-5-1">
        <title>Pre-coordinated concepts categories:</title>
        <p>the content model5</p>
      </sec>
      <sec id="sec-5-2">
        <title>Postcoordinated Categories :CAST approach</title>
        <p>
          To try to disambiguate this blend of terminology and
information model a CAST approach was developed
          <xref ref-type="bibr" rid="ref17">(Rodrigues,
2009)</xref>
          following different use cases.
        </p>
        <p>The prerequisite was that diseases, disorders, injuries, signs
and symptoms (all of which matter in ICD) cannot be
attributed to one ontological category either.</p>
        <p>
          The Ontology for General Medical Sciences (OGMS)
          <xref ref-type="bibr" rid="ref18">(Scheuermann, 2009)</xref>
          made the attempt to interpret diseases
as dispositions and disorders at pathological body structures,
both distinguished from disease courses as processes. This
approach was criticized by
          <xref ref-type="bibr" rid="ref12">(Schulz, 2011)</xref>
          who argued that
this overloads the meaning of intrinsically shallow terms
which are often used as synonyms. Both groups, however,
agreed on the observation that there are many terms for
which different meanings co-exist: “allergy” may denote an
allergic disposition or an allergic manifestation, “tumor” a
lump of tissue or a process, and “fracture” a broken bone or
an injury and healing process.
        </p>
        <p>Three different CAST were proposed for Disease, Diagnosis
and Patient Findings and Problems
Disease is the most complete view. It is based on an
abnormality in the Body System OR Structure (morphology) or in
the Functional properties (patho-physiology), Causal
properties and Manifestation Properties.</p>
        <p>Disease/disorder furthermore has to be distinguished from
Diagnosis, which represents the view of a clinical decision
5 www.who.int/classifications/icd/revision/contentmodel/en/index.html
maker who shall take decisions in sometimes uncertain
situations. Diagnoses are based on a set of Manifestation
properties and Diagnostic criteria to be defined by the
clinical domain-specific ICD-11 revision Topic Advisory
Groups TAGs. Diagnoses principally include the possibility
that they do not correspond to any pathological entity in the
patient (suspected diagnoses or false diagnoses).</p>
        <p>Finally Patient Findings and Problems (signs, symptoms,
syndromes, test results, situations, etc.) are very often
mentioned in health record for surveillance or other without
reaching the level of a diagnosis assumption.
4.3</p>
      </sec>
      <sec id="sec-5-3">
        <title>The Semantic Links</title>
        <p>They are the relations has_finding, has_location,
has_abnormality, has_etiology and has_focus.
 has_finding is the link authorised between Disease or</p>
        <p>Diagnosis and Manifestation Properties.
 has_location is the link authorised between Disease or</p>
        <p>Diagnosis and Body System/structure.
 has_abnormality is the link authorised between
Disease and Functioning Properties or Body
System/structure /Morphology.
 has_etiology is the link authorised between Disease or</p>
        <p>Diagnosis and Causal Properties
 has_focus is the link authorised between Patient
Findings and Problems and Manifestation Properties,
BodySystem/structure, Functioning Properties.
4.4</p>
      </sec>
      <sec id="sec-5-4">
        <title>The Minimal Domain Constraints</title>
        <p>Patient Findings and Problems: at least one has_focus.
Diagnosis: at least one has_finding and at least one
has_location
Disease: at least one has_finding, at least one has_location,
at least one has_abnormality and at least one has_etiology.
5</p>
      </sec>
    </sec>
    <sec id="sec-6">
      <title>DISCUSSION</title>
      <p>We presented the CAST approach on different healthcare
sub domains (Surgical Procedures, Patient Safety, ICD 11)
and the relations between this approach and ULO
approaches (BFO and Bio Top).</p>
      <p>We showed in a first step that CAST is challenging on one
hand basic ontological entities and on the other the relations
(Bio Top or BFO and RO).</p>
      <p>
        In the following part we demonstrated that the use of the
CAST approach for the ICD-11 revision is facing the
ambiguity and fuzziness of the healthcare terminologies which
are oversimplifying the knowledge they contain to ease their
utilization. ICD 11 CAST blends ontology with information
model (and therefore epistemic) aspects, as in PS CAST.
Recently an ontological triad structure-named
structuredisposition process (SDP) has been proposed to
disambiguate complex pathological process in another healthcare
terminology SNOMED CT and the creation of a BioTop
disjunctive class recommended
        <xref ref-type="bibr" rid="ref12">(Schulz, 2011)</xref>
        .
A preliminary analysis has shown the difficulties and
possible solutions (e.g. to use dispositions), which can still be
embedded into a consistent ontological framework.
Our next steps consist in the completion of DL models of
both PS-CAST and ICD11-CAST, using BioTop,
disjunctive entity and, in a further stage, BFO version 2.
This can contribute to the WHO/IHTSDO Joint Advisory
Group (JAG) works on a common ontological basis for the
Foundation Component (FC) of ICD 11 and SNOMED CT.
In another field it would be useful that ontological
alignment should become a routine part of CEN/ISO CAST
development, for it can be exploited to discover similarities
and differences between the existing CAST suite found in
CEN/ISO publications.
      </p>
      <p>The empirical nature of the CAST approach influences what
should be represented - if some of these requirements are
epistemic then the CAST should support them. If principled
ontological frameworks cannot accommodate epistemic
aspects this is not a reason to marginalize or reject them
from the models (and associated knowledge products).
The CAST, BFO and BioTop approaches are truly
complementary. They are providing an empirical 'reality check' to
quality improvement strategies that may disallow the
inclusion of epistemic notions. They can support the process of
ontology construction and ease ontology application when
complex healthcare knowledge is concerned.</p>
      <p>The crucial value in having computable models of these
terminologies lies in the use of formal reasoning
mechanisms for validation. Built into the workflow of terminology
construction and maintenance, such a combined CAST ULO
approach would be an important asset towards high-quality
biomedical terminological systems.</p>
    </sec>
    <sec id="sec-7">
      <title>ACKNOWLEDGEMENTS</title>
      <p>PS-CAST project has been contracted by WHO
(Registration #2009/33635-0, Order 200094768, Reg. File H15 APW
221) as part of ICPS development.</p>
    </sec>
  </body>
  <back>
    <ref-list>
      <ref id="ref1">
        <mixed-citation>
          <string-name>
            <surname>Rodrigues</surname>
            <given-names>JM</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Kumar</surname>
            <given-names>A</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Bousquet</surname>
            <given-names>C</given-names>
          </string-name>
          ,
          <article-title>Trombert B</article-title>
          .
          <article-title>Standards and biomedical terminologies: the CEN TC 251 and ISO TC 215 categorial structures. A step towards increased interoperability</article-title>
          .
          <source>Stud Health Technol Inform</source>
          .
          <year>2008</year>
          ;
          <volume>136</volume>
          :
          <fpage>857</fpage>
          -
          <lpage>62</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref2">
        <mixed-citation>
          <string-name>
            <surname>Rector</surname>
            <given-names>AL</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Solomon</surname>
            <given-names>WD</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Nowlan</surname>
            <given-names>WA</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Rush</surname>
            <given-names>TW</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Zanstra</surname>
            <given-names>PE</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Claassen</surname>
            <given-names>WM</given-names>
          </string-name>
          .
          <article-title>A Terminology Server for medical language and medical information systems</article-title>
          .
          <source>Methods Inf Med</source>
          <year>1995</year>
          ;
          <volume>34</volume>
          (
          <issue>1-2</issue>
          ):
          <fpage>147</fpage>
          -
          <lpage>57</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref3">
        <mixed-citation>
          <string-name>
            <surname>Baader</surname>
            <given-names>F</given-names>
          </string-name>
          et al. (
          <year>2007</year>
          ).
          <article-title>The Description Logic Handbook. 2nd edition</article-title>
          . Theory, Implementation, and
          <string-name>
            <surname>Applications</surname>
          </string-name>
          , Cambridge, U.K.: Cambridge University Press.
        </mixed-citation>
      </ref>
      <ref id="ref4">
        <mixed-citation>
          <string-name>
            <surname>Horridge</surname>
            <given-names>M</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Patel-Schneider</surname>
            <given-names>P</given-names>
          </string-name>
          (
          <year>2009</year>
          ).
          <article-title>OWL 2 Web Ontology Language Manchester Syntax</article-title>
          . http://www.w3.org/TR/2009/NOTE-owl2
          <string-name>
            <surname>-</surname>
          </string-name>
          manchester-syntax-
          <volume>20091027</volume>
          /
        </mixed-citation>
      </ref>
      <ref id="ref5">
        <mixed-citation>
          <string-name>
            <surname>Smith</surname>
            <given-names>B</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Grenon</surname>
            <given-names>P</given-names>
          </string-name>
          : Basic Formal Ontology.
          <year>2011</year>
          [www.ifomis.org/bfo],
          <source>Last accessed: Jan</source>
          <volume>30</volume>
          ,
          <year>2012</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref6">
        <mixed-citation>
          <string-name>
            <surname>Beißwanger</surname>
            <given-names>E</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Schulz</surname>
            <given-names>S</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Stenzhorn</surname>
            <given-names>H</given-names>
          </string-name>
          and
          <string-name>
            <surname>Hahn U. BioTop</surname>
          </string-name>
          <article-title>: An Upper Domain Ontology for the Life Sciences - A Description of its Current Structure, Contents, and Interfaces to OBO Ontologies</article-title>
          .
          <source>Applied Ontology</source>
          <year>2008</year>
          ;
          <volume>3</volume>
          ;4:
          <fpage>205</fpage>
          -
          <lpage>212</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref7">
        <mixed-citation>
          <string-name>
            <surname>Rodrigues</surname>
            <given-names>JM</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Casey</surname>
            <given-names>A</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Bousquet</surname>
            <given-names>C</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Kumar</surname>
            <given-names>A</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Lewalle</surname>
            <given-names>P</given-names>
          </string-name>
          , Trombert Paviot B.
          <article-title>Comparing existing national and international classification systems of surgical procedures with the CEN/ISO 1828 ontology framework standard Stud Health Technol Inform</article-title>
          .
          <year>2011</year>
          ;
          <volume>169</volume>
          :
          <fpage>844</fpage>
          -
          <lpage>8</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref8">
        <mixed-citation>
          <string-name>
            <surname>Smith</surname>
            <given-names>B</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Ceusters</surname>
            <given-names>W</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Klagges</surname>
            <given-names>B</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Köhler</surname>
            <given-names>J</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Kumar</surname>
            <given-names>A</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Lomax</surname>
            <given-names>J</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Mungall</surname>
            <given-names>C</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Neuhaus</surname>
            <given-names>F</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Rector</surname>
            <given-names>AL</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Rosse</surname>
            <given-names>C.</given-names>
          </string-name>
          <article-title>Relations in biomedical ontologies</article-title>
          .
          <source>Genome Biol</source>
          .
          <year>2005</year>
          ;
          <volume>6</volume>
          (
          <issue>5</issue>
          ):
          <fpage>R46</fpage>
        </mixed-citation>
      </ref>
      <ref id="ref9">
        <mixed-citation>
          <string-name>
            <surname>Gangemi</surname>
            <given-names>A</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Guarino</surname>
            <given-names>N</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Masolo</surname>
            <given-names>C</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Oltramari</surname>
            <given-names>A</given-names>
          </string-name>
          and
          <string-name>
            <surname>Schneider L. Sweetening</surname>
          </string-name>
          <article-title>Ontologies with DOLCE</article-title>
          .
          <source>Lecture Notes in Computer Science</source>
          ,
          <year>2002</year>
          , Volume
          <volume>2473</volume>
          /
          <year>2002</year>
          ,
          <fpage>223</fpage>
          -
          <lpage>233</lpage>
          , DOI: 10.1007/3-540-45810-7_
          <fpage>18</fpage>
        </mixed-citation>
      </ref>
      <ref id="ref10">
        <mixed-citation>
          <string-name>
            <surname>Runciman</surname>
            <given-names>W</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Hibbert</surname>
            <given-names>P</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Thomson</surname>
            <given-names>R</given-names>
          </string-name>
          , Van Der Schaaf
          <string-name>
            <given-names>T</given-names>
            ,
            <surname>Sherman</surname>
          </string-name>
          <string-name>
            <given-names>H</given-names>
            ,
            <surname>Lewalle</surname>
          </string-name>
          <string-name>
            <surname>P</surname>
          </string-name>
          .
          <article-title>Towards an International Classification for Patient Safety: key concepts and terms</article-title>
          ,
          <source>Int J Qual Health Care</source>
          .
          <year>2009</year>
          ;
          <volume>21</volume>
          (
          <issue>1</issue>
          ):
          <fpage>18</fpage>
          -
          <lpage>26</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref11">
        <mixed-citation>
          <string-name>
            <surname>Souvignet</surname>
            <given-names>J</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Bousquet</surname>
            <given-names>C</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Lewalle</surname>
            <given-names>P</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Trombert-Paviot</surname>
            <given-names>B</given-names>
          </string-name>
          and
          <string-name>
            <surname>Rodrigues JM</surname>
          </string-name>
          .
          <article-title>Modeling patient safety incidents knowledge with the categorial structure method</article-title>
          .
          <source>AMIA Annu Symp Proc</source>
          .
          <year>2011</year>
          ;
          <year>2011</year>
          :
          <fpage>1300</fpage>
          -
          <lpage>8</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref12">
        <mixed-citation>
          <string-name>
            <surname>Schulz</surname>
            <given-names>S</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Spackman</surname>
            <given-names>K</given-names>
          </string-name>
          ,
          <string-name>
            <surname>James</surname>
            <given-names>A</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Cocos</surname>
            <given-names>C</given-names>
          </string-name>
          and
          <string-name>
            <surname>Boeker M.</surname>
          </string-name>
          <article-title>Scalable representations of diseases in biomedical ontologies</article-title>
          .
          <source>Journal of Biomedical Semantics</source>
          <year>2011</year>
          ,
          <volume>2</volume>
          (
          <issue>Suppl</issue>
          .2):S6 doi:10.1186/2041-1480-2-S2-S6
        </mixed-citation>
      </ref>
      <ref id="ref13">
        <mixed-citation>
          World Health Organization.
          <article-title>International Classification of Diseases (ICD). www</article-title>
          .who.int/classifications/icd/.
          <source>Last accessed: Jan</source>
          <year>2012</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref14">
        <mixed-citation>
          <string-name>
            <surname>Tudorache</surname>
            <given-names>T</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Falconer</surname>
            <given-names>S</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Nyulas</surname>
            <given-names>C</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Storey</surname>
            <given-names>MA</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Ustün</surname>
            <given-names>TB</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Musen</surname>
            <given-names>MA</given-names>
          </string-name>
          .
          <article-title>Supporting the Collaborative Authoring of ICD-11 with WebProtégé</article-title>
          .
          <source>Proceedings of the AMIA Annual Symposium</source>
          <year>2010</year>
          ,
          <fpage>802</fpage>
          -
          <lpage>806</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref15">
        <mixed-citation>
          <string-name>
            <given-names>Rodrigues</given-names>
            <surname>JM. ICD-</surname>
          </string-name>
          <article-title>11 revision: where are we now? Ontology-driven tools and the web platform</article-title>
          .
          <source>BMC Health Services Research</source>
          <year>2010</year>
          ,
          <volume>10</volume>
          (
          <issue>Suppl 2</issue>
          ):A7 doi:10.1186/
          <fpage>1472</fpage>
          -6963-10-S2-A7
        </mixed-citation>
      </ref>
      <ref id="ref16">
        <mixed-citation>
          <string-name>
            <given-names>S. W.</given-names>
            <surname>Tu</surname>
          </string-name>
          , et al.
          <article-title>A content model for the ICD-11 revision</article-title>
          .
          <source>Technical Report BMIR-2010-1405</source>
          , Stanford Center for Biomedical Informatics Research,
          <year>2010</year>
          .
        </mixed-citation>
      </ref>
      <ref id="ref17">
        <mixed-citation>
          <string-name>
            <surname>Rodrigues</surname>
            <given-names>JM</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Kumar</surname>
            <given-names>A</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Bousquet</surname>
            <given-names>C</given-names>
          </string-name>
          ,
          <article-title>Trombert B</article-title>
          .
          <article-title>Using the CEN/ISO standard for categorial structure to harmonise the development of WHO international terminologies</article-title>
          .
          <source>Studies of Health Technology and Informatics</source>
          <year>2009</year>
          ;
          <volume>150</volume>
          :
          <fpage>255</fpage>
          -
          <lpage>259</lpage>
          .
        </mixed-citation>
      </ref>
      <ref id="ref18">
        <mixed-citation>
          <string-name>
            <surname>Scheuermann</surname>
            <given-names>RH</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Ceusters</surname>
            <given-names>W</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Smith</surname>
            <given-names>B</given-names>
          </string-name>
          :
          <article-title>Toward an Ontological Treatment of Disease and Diagnosis</article-title>
          .
          <source>In Proc. of the 2009 AMIA Summit on Translational Bioinformatics</source>
          . San Francisco, California. American Medical Informatics Association;
          <year>2009</year>
          :
          <fpage>116</fpage>
          -
          <lpage>120</lpage>
          .
        </mixed-citation>
      </ref>
    </ref-list>
  </back>
</article>