An Ontology driven collaborative development for biomedical terminologies: from the French CCAM to the Australian ICHI coding system

1
An Ontology driven collaborative development for
biomedical terminologies from the French CCAM to
the Australian ICHI coding system

• a University of Saint Etienne France,
• b University of Manchester, United Kingdom
• c Radboud University of Nijmegen Medical Centre
  ,The Netherlands
• d University Hospitals of Geneva, Switzerland
• e National Centre for Classification in Health,
  Australia
• f Freiburg University Hospital, Germany
• g Nova University of Lisbon, Portugal
• rodrigues_at_univ-st-etienne.fr

2
(No Transcript)
3
Terminology to Ontology

• 1 Introduction
• 2 Definitions
• 3 Galen and CCAM
• 4 ICHI Galen
• 5 Discussion
• 6 Conclusion

4
Terminology to Ontology

• 1 Introduction
• 2 Definitions
• 3 Galen and CCAM
• 4 ICHI Galen
• 5 Discussion
• 6 Conclusion

5
ACRONYMS.

• GALEN (Generalised Architecture for Languages,
  Encyclopaedias and Nomenclatures in Medicine) EU
  project
• ICHI (International Classification of Health
  Interventions)NCCH Australia
• CCAM (Classification Commune des Actes Médicaux)
  France
• CEN standards on categorial structure
• EN 12 264 Categorial structure
• EN 1828 Categorial structure for surgical
  interventions

6
TERMINOLOGY

• Term meaning
• Terminology artefacts ( nomenclature ,flat list
  of controlled vocabularies ,classifications and
  thesaurus,coding system)
• Used to analyse the reality from instances of
  universals (entities invariant in time and space)
  known or to know
• Used by humans with flexibility and ambiguity

7
FIRST GENERATION TERMINOLOGICAL SYSTEMS 1
FIRST GENERATION
presentation systematic list(typically with
codes) , alphabetical index with permutations
organization fixed (typically one,
hierarchical) purposes devoted to a single
application flexibility NO extension
restricted to author processing on semantics
NO Examples ICD9, ICD9 CM, ICD10 ICPM OPCS
CDAM ICHI
8
SECOND GENERATION TERMINOLOGICAL SYSTEMS 2
SECOND GENERATION
presentation categorial structure
cross-thesaurus lists (classifications,
nomenclatures, ...) organization dynamic
(multiple, hierarchical) purposes multiple
flexibility and extensions Yes with
restrictions new combinations can be made by
users but without knowledge validation
processing on semantics Yes but restricted to
the predefined categorial structure Examples
SNOMED III, ICD10 PCS ,CCI,CCAM LOINC-LAB ,
DICOM-SDM (Snomed Dicom Microglossary
imaging) MCTGE Minimum Common Terminology for
Gastro intestinal Endoscopy CISP/ICPC
,Braun, CIF,DSMIII and DSMIV
9
THIRD GENERATION TERMINOLOGICAL SYSTEMS 3
THIRD GENERATION
presentation categorial structure
cross-thesaurus lists knowledge based model
/ ontology model (all and only sensible medical
concepts) organization dynamic (multiple,
hierarchical) purposes multiple
flexibility and extensions unrestricted new
combinations validated by knowledge base model
(by predefined combinatorial rules) processing
on semantics Yes complete (formal
processing) Examples SNOMED-RT and
CT? GALEN FMA GO

10
TERMINOLOGY and ONTOLOGY

• Third generation terminologies are a set of
  terminological artefacts related to a knowledge
  representation of all and only sensible
  concepts-universals in a compositional structure
  allowing multiple hierarchies this
  representation being a type of ontology .

11
ONTOLOGY vs TERMINOLOGY

• Ontology a representation of some pre-existing
  domain of reality
• (1) reflects the properties of the entities
  within its domain in such a way that it obtains
  a systematic correlation between reality and the
  representation itself and the non ambiguous
  comparison between different representations of
  the reality
• (2) is intelligible to a domain expert,
• (3) is formalised in a way that allows it to
  support automatic information processing .
• Distinct from Terminology

12
ONTOLOGY vs TERMINOLOGY

• 3 types for ontology
• 1 Top level/Meta Ontology
• 2 Domain Reference Ontology
• Ontology for anatomy
• Ontology for biology
• Ontology for interventions
• 3 Application Ontology
• Related to clinical terminologies
• Galen 1 et 2
• FMA 2
• GALEN-CCAM ICHI and CEN categorial structures 3 ?

13
Terminology to Ontology

• 1 Introduction
• 2 Definitions
• 3 Galen and CCAM
• 4 ICHI Galen
• 5 Discussion
• 6 Conclusion

14
Galen model

• A two steps approach  .
• 1Top down modelling by
• -full categorisation of real world
• -multiple inheritance composite concepts(meaning)
  hierarchy
• -semantic relations between them
• -sanctions tests constraining the composition of
  concepts
• Formal language GRAIL (Galen Representation and
  Integration Language).
• 2 Bottom up extension
• -pragmatic reverse engineering
• -surgical interventions GALEN-In-USE
• -centres in Belgium,Finland,France,Germany,
  Greece,Italy,The Netherlands,UK , Sweden
• - diagnostic Hungary
• 9400 sensible tests

15
GALEN CORE Model top level category space
16
Extended GALEN scheme for Surgical Procedure
(GASP)

17
CCAM-developments

• CCAM development
• Domain experts consensus in French language (1.5
  person-century)
• GALEN tools application
• Semi formal representation by junior physicians
• Mapping to GRAIL language GALEN formal
  representation
• Natural language processing generation to French
  language
• Conformance to European standards EN 1828 EN12
  264
• GALEN representation for 7478 surgical
  procedures
• 700 new concepts
• Using 2400 descriptors and 59 semantic links

18
CCAM- GALEN - representation

• 7478 rubrics of composite surgical procedures
  (X3)
• 2400 concept descriptors and 59 semantic links
• 2400 descriptors,
• 1297 descriptors for Anatomy category with 30
  semantic categories
• 271 descriptors for Pathology category
• 231 descriptors for Device category
• 186 descriptors for Deed category
• 59 semantic links
• 27 concerning 4 categories
• 7 only for Pathology (motivated by ,to achieve
  et..)

19
Linguistic and Ontology Assessments (adapted from
Baud and Rassinoux)
GALEN expression
GeneralisedProcess- (playsClinicalRole)-gtS
urgicalRole (isMainlyCharacterisedBy)-gtperf
ormance- (isEnactmentOf)-gtExcising-
(hasSpecificSubprocess)-gtAccessing
- (hasSpecificSubprocess)-gtIn
cising-
(actsSpecificallyOn)-gtSkull
(playsClinicalRole)-gtSurgicalRole\
(playsClinicalRole)-gtSurgicalRole\
(actsSpecificallyOn)-gtBodyCavity-
(isSpaceDefinedBy) TruelyHoll
owBodyStructure(isContainedIn)-gtIntracranialSpace
(contains)-gtPus
(hasPathologicalStatus)-gtpathological\
(playsClinicalRole)-gtSurgicalRole\\\.
GENERATOR
CCAM assessment
Compositional modeling
Linguisticevaluation
Dissection
20
Step 1 MAKE AN INTERMEDIATE DISSECTION(claw )

• 1) RUBRIQUE INITIALE U2.050
• Résection endoscopique cervico-prostatique, avec
  urétrotomie à laveugle, urétrocystoscopie et
  mise en place d'une sonde urétrale.
• 2) DISSECTION INTERMEDIAIRE
• RUBRIC "Résection endoscopique cervico-prostatique
  , avec urétrotomie à l'aveugle,
  urétrocystoscopie et mise en place d'une sonde
  urétrale"
• SOURCE USE CODE "U2.050"
• MAIN resecting
• ACTS_ON Anatomy neck_of_urinary_bladder
  prostate
• BY_MEANS_OF endoscope
• BY_TECHNIQUE inspecting
• ACTS_ON Anatomy urinary bladder urethra
• BY_MEANS_OF Device endoscope
• WITH_GUIDANCE_BY Device endoscope
• WITH installing
• ACTS_ON Device catheter
• HAS_DESTINATION Anatomy urethra
• WITH incising
• ACTS_ON Anatomy urethra
• HAS_APPROACH closed approach

21
Step 2Expand to formal Representation
3) GRAIL REPRESENTATION cl_SurgicalDeed,rel_
isMainlyCharacterisedBy(cl_performance,rel_isEna
ctmentOf(cl_NLGCPT_Resecting,rel_actsSpecificall
yOn(cl_ArbitraryBodyConstruct,rel_hasStructuralC
omponent(cl_UrinaryBladderNeck),rel_hasStructuralC
omponent(cl_ProstateGland)),rel_hasSpecificPhysi
calMeans(cl_Endoscope),rel_hasSpecificSubprocess(
cl_Inspecting,rel_actsSpecificallyOn(cl_Arbitrar
yBodyConstruct,rel_hasStructuralComponent(cl_Urin
aryBladder),rel_hasStructuralComponent(cl_Urethra)
),rel_hasSpecificPhysicalMeans(cl_Endoscope))
)),rel_isCharacterisedBy(cl_performance,rel_i
sEnactmentOf(cl_NLGCPT_InstallingProcess,rel_Loc
ativeAttribute(cl_Catheter,rel_isActedOnSpecific
allyBy(cl_Transport,rel_hasSpecificConsequence(
cl_Displacement,rel_hasBetaConnection(cl_Urethra)
))))),rel_isCharacterisedBy(cl_performa
nce,rel_isEnactmentOf(cl_NLGCPT_Incising,rel_ac
tsSpecificallyOn(cl_Urethra),rel_hasSpecificSubpro
cess(cl_SurgicalApproaching,rel_hasSurgicalOpenC
losedness(cl_SurgicalOpenClosedness,rel_hasAbsol
uteState(cl_surgicallyClosed)))))
22
Step 3Controlled natural language generation
(see S Hansen Freiburg)
Original résection endoscopique
cervico-prostatique, avec urétrotomie à
laveugle, urétrocystoscopie et mise en place
d'une sonde urétrale Génération sans contraction
acte chirurgical de résection endoscopique
d'une entité corporelle ayant pour partie le col
de la vessie et ayant pour partie la prostate
sous inspection endoscopique d'une entité
corporelle ayant pour partie la vessie et ayant
pour partie l'urètre avec mise en place d'un
cathéter dans l'urètre et avec incision urétrale
et fermée. Génération avec contraction
résection sous endoscope du col de la vessie et
de la prostate avec inspection endoscopique de
lurètre et de la vessie avec incision fermée de
lurètre et mise en place dun cathéter dans
lurètre
23

QA with ontology driven tools

• What is an arthroplasty ?
• installing a prosthesis within the joint to
  replace at least a part of it
• R0232b Arthroplastie du poignet par prothèse
  partielle, à ciel ouvert
• R0847 Arthroplastie inter somatique par prothèse
  discale
• R1285b Reconstruction de l'articulation du genou
  par prothèse massive ou sur mesure, après perte
  de substance segmentaire.Avec ou sans
  reconstruction de l'appareil extenseur
• R0064 Arthroplastie scapulo humérale par prothèse
  humérale unipolaire ou par cupule mobile
• R0231 Arthroplastie radio carpienne par prothèse
  totale à ciel ouvert
• R0403 Arthroplastie coxo-fémorale par cupule
  fémorale

24

QA with ontology driven tools

• What is an arthroplasty ?
• installing a prosthesis within the joint to
  replace at least a part of it or
• A plastic repairing of the components of a joint
• Without installing a prosthesis to replace the
  joint?
• R0290b Arthroplastie de plusieurs articulations
  d'un rayon de la main
• R0289b Arthroplastie d'une articulation d'un
  rayon de la main.A l'exclusion de arthroplastie
  de l'articulation carpo-métacarpienne du pouce
• R0233 Arthroplastie de larticulation
  carpo-métacarpienne du pouce avec ou sans plastie
  de lappareil capsulo ligamentaire

25
Terminology to Ontology

• 1 Introduction
• 2 Definitions
• 3 Galen and CCAM
• 4 ICHI Galen
• 5 Discussion
• 6 Conclusion

26
ICHI GALEN METHOD

• Semi formal representation using SPET (Kermanog
  Inc) by
• University of Saint Etienne (France)
• Radboud University Nijmegen (The Netherlands)
• Mapping to GRAIL language GALEN formal
  representation
• University of Manchester (UK)
• Natural Language Processing generation to
  English, Portuguese and French languages
• University Hospitals of Geneva (Switzerland)
• University of Freiburg (Germany)
• Conformity to European Standards EN 1828 and EN
  12264

27
QA with ontology driven toolsWhat is an
arthroplasty ?A plastic repairing of the
components of a joint without installing a
prosthesis to replace the joint?Code 1518
initial rubric ICHI-Chapter XV,1518
Arthroplasty of knee generated label En
Repairing of the knee joint . PoReparação da
articulacâo do joelho. Fr Réparation de
larticulation du genou.
28
QA with ontology driven tools What is an
arthroplasty ?installing a prosthesis within the
joint to replace at least a part of it code 1519
initial rubric ICHI-Chapter XV,1519
Arthroplasty of knee with bone graft to femur or
tibia generated label En Replacement of the
knee joint with a prosthesis of the knee
joint,with installation of bone graft tissue in
the femur or the tibia by means of harvesting
bone . PoSubstituição da articulacâo do joelho
cum una prótese da articulacâo do joelho, cum
instalacão de transplante de tecido de osso no
fémur ou na tíbia por meio de colhida do
osso. Fr Remplacement de larticulation du genou
par une prothèse de larticulation du genou avec
mise en place de tissu osseux greffé sur le fémur
ou le tibia au moyen de prélèvement osseux.
29
QA with ontology driven tools What is an
arthroplasty ?Code 1589
initial rubric ICHI-Chapter XV,1589
Arthroplasty of hip joint generated label En
Repairing of the hip joint . Plastic repairing
or Hip replacement with a prosthesis? PoReparaçã
o da articulacâo do anca/quadril. Reparação
plástico ou Substituição da articulacâo do anca
cum una prótese ? Fr Réparation de
larticulation de la hanche. Réparation plastique
ou Prothèse de hanche?
30
Terminology to Ontology

• 1 Introduction
• 2 Definitions
• 3 Galen and CCAM
• 4 ICHI Galen
• 5 Discussion
• 6 Conclusion

31
GALEN-CCAM lessons (coordination terminology
ontology)

• Coordination of terminology and ontology has
  allowed participation of clinical experts
  colleges in their natural language (bottom up)
• Has Mixed contradictory trends
• Terminology artefacts aligned to evidence rather
  than God belief
• Supporting contradiction
• To allow linguistic ambiguity monitoring for
  humans
• Description logic is necessary for
  machinecomputers
• Different end users goals need different
  terminology artefacts

32
ICHI GALEN

• 3 goals to 4
• - To demonstrate that a compositional approach to
  developing WHO candidate terminologies is
  feasible and cost-effective
• - To demonstrate a more collaborative approach to
  building international terminologies, including
  open developments (OpenGalen)
• -   To demonstrate that automatic translations
  based on an ontology make the translation process
  more transparent faster and of better quality
• -   To provide evidence on comparative costs with
  the traditional domain experts method to inform
  future decision making ?

33
Terminology to Ontology

• 1 Introduction
• 2 Definitions
• 3 Galen and CCAM
• 4 ICHI Galen
• 5 Discussion
• 6 Conclusion

34
CONCLUSION 1

• This work has demonstrated the feasibility to
  apply ontology driven tools designed for a
  national coding system (French CCAM) within a EU
  funded project and a collaborative approach to
  the development of a terminology ICHI candidate
  to become the WHO classification of health
  interventions .
• Such a method has shown 20 inconsistencies in
  traditional coding systems and convinced
  international and national organisations that it
  was a better and easier method to maintain and to
  share coding systems between countries than the
  traditional domain experts consensus method used
  until now.

35
CONCLUSION

• It is still necessary to convince the decision
  makers that it can be faster and cheaper by
  multiplying the number of modelling centres
  working in their local language and investing in
  few web based tools platforms than in the
  traditional bottomless domain expertise.
• As the healthcare systems are becoming more and
  more knowledge-based and communicating the
  terminology they use must be able to allow their
  expressiveness and in a sense their ambiguity but
  must be constrained by a formal logic
  representation or ontology based within the
  computer to insure the safety of healthcare
  delivery.
• The process we have developed for classifications
  and coding systems shall be tested for any
  clinical terminology (Snomed CT?).

36
ENDE Vielen Dank für Ihre Aufmerksamkeit Thank
you Merci Gracie Gracias Oblig
ado Efcharisto ARIGATO GOZAIMASU Tag M
ultumesc.