HAL Id: hal-01212095
https://hal.archives-ouvertes.fr/hal-01212095
Submitted on 6 Oct 2015
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Knowledge-based modelling applied to synucleinopathies
Bernard Kamsu-Foguem, Pierre Tiako, Edward Mutafungwa, Clovis Foguem
To cite this version:
Bernard Kamsu-Foguem, Pierre Tiako, Edward Mutafungwa, Clovis Foguem. Knowledge-based mod- elling applied to synucleinopathies. European Geriatric Medicine, Elsevier, 2015, 6 (n° 6), pp. 381-388.
�10.1016/j.eurger.2015.02.011�. �hal-01212095�
To link to this article: DOI:10.1016/j.eurger.2015.02.011 http://dx.doi.org/10.1016/j.eurger.2015.02.011
This is an author-deposited version published in: http://oatao.univ-toulouse.fr/
Eprints ID: 14090
To cite this version:
Kamsu-Foguem, Bernard and Tiako, Pierre and Mutafungwa, Edward and Foguem, Clovis Knowledge-based modelling applied to synucleinopathies.
European Geriatric Medicine. ISSN 1878-7649
Open Archive Toulouse Archive Ouverte (OATAO)
OATAO is an open access repository that collects the work of Toulouse researchers and makes it freely available over the web where possible.
Any correspondence concerning this service should be sent to the repository administrator: staff-oatao@listes-diff.inp-toulouse.fr
Technologyappliedtogeriatricmedicine
Knowledge-based modelling applied to synucleinopathies
B.Kamsu-Foguema,*,P.F.Tiakob,E. Mutafungwac,C.Foguemd
aLaboratoryofProductionEngineering(LGP),EA1905,ENIT-INPT,UniversityofToulouse,47,avenued’Azereix,BP1629,65016Tarbescedex,France
bCenterforITResearch,LangstonUniversity,OK73050,USA
cAaltoUniversitySchoolofElectricalEngineering,PL13000,00076Aalto,Espoo,Finland
dCenterforFoodandTastesciences(CSGA)–UMR6265CNRS–UMR1324INRA–UniversityofBurgundy,9E,boulevardJeanne-d’Arc,21000Dijon,France
1. Introduction
Thecurrentmedicaldemographyanditsunevendistribution, particularlyin somemedical disciplines,arecreatingsignificant challengesinensuringcontinuityof care.Thishasresultedina pressingneedformedicalfacilitiesinplaceswheretheshortageof specialistsisacute.Telemedicineisatoolthatenablesequalaccess tomedicalexpertiseregardlessofplaceofcare[1].Thispractice allows physicians of regional institutions to ensure access to competencefor differentmedical investigationsorcasesand to benefit from the expertise of their colleagues from different specializations (radiology, surgery, psychiatry or neurology) [2].Indeed,telemedicinehasalloweddoctorshospitalsandclinics tohavea network for accessingremoteresources andcomple- mentaryskills[3].Otherbenefitsoftelemedicineincludepatients whoarelesstiredandstressedandfamilieswhonolongerhaveto travellongdistancestovisittheirdoctors[4].Medicalpractitioners now have in their offices or in their local hospitals capacity identicaltothoseatmajorhospitals.Theseadvancesarepotential safeguard measures to ensure the protection of some local
hospitals. In summary, telemedicine provides impact in three keydimensions[5]:
medical impact through improved medical diagnosis and treatment,aswellas,theincreasedcompetenceofpractitioners participatinginthenetwork;
territorialimpactbymaintainingskillsandspecializedservices inhospitalsandclinicsinmediumandsmallercities,especially inruralareas;
medico-economicimpactthrough reductioninthenumber of unnecessarytransfersand theincreaseinthenumberof vital transferswithausefulgainforpatientswhilecontrollingcosts.
Theeraoftechnologicalnetworksinmedicineisinevitableto improve the quality, safety, and continuity of care, while promotingequalaccesstohighqualitylocalcarebypromoting coordinationandcooperationofallstakeholders.Atthetechno- logical level, further progress can be made in the field of telemedicineifthemainobstaclesareovercome[6].Bydeploying this technology in all remote hospitals or care centers, one improvesconnectivitybysettingapplicationofqualitystandards andbuildingarelationshipoftrust.Telemedicineefficiencyshould allow equitable management of patients. It allows sharing of Keywords:
Telemedicine Ontologicalknowledge Collaboration Semanticmodelling Geriatrics
ABSTRACT
Theadoptionoftelemedicinetechnologieshasenabledcollaborativeprogramsinvolvingavarietyof links among distributed medical structures and health officials and professionals. The use for telemedicinefortransmissionofmedicaldataandthepossibilityforseveraldistantphysicianstoshare theirknowledge on given medical cases provides clear benefits,but also raises severalunsolved conceptualandtechnicalchallenges.Theseamlessexchangeandaccessofmedicalinformationbetween medicalstructures,healthprofessionals,andpatientsisaprerequisitefortheharmoniousdevelopment ofthisnewmedicalpractice.Thispaper proposesanewapproachofsemanticinteroperabilityfor enabling mutual understanding of terminologies and concepts used. The proposed semantic interoperabilityapproachisbasedonconceptualgraphtosupportcollaborativeactivitiesbydescribing howdifferent healthspecialists can applyappropriatestrategies to eliminate differential medical diagnosis.Intelligentanalysisstrategiesareusedtonarrowdownandpinpointmedicaldisorders.The modelproposedisfullyverifiedbyacasestudyinthecontextofelderlypatientsandspecificallydealing withsynucleinopathies,agroupofneurodegenerativediseasesthatincludeParkinson’sdisease(PD), dementiawithLewybodies(DLB),pureautonomicfailure(PAF)andmultiplesystematrophy(MSA).
* Correspondingauthor.Tel.:+33624302337;fax:+33562442708.
E-mailaddress:Bernard.Kamsu-Foguem@enit.fr(B.Kamsu-Foguem).
qualityimageviewingstations(adapteddiagnostic),flexibleusage ofresources(possibilityofWebaccess),and,toensurethat the benefitsof suchtelemedicine canbemutual, providesopportu- nities for technology transfer and improved efficiency in the healthcare sectors of partners institutions.The development of telemedicineisbasedonstrengtheningourinnovationcapabilities andrequiressustainedcooperationthatcutsacrossthefieldsof health, industry, research, and defence [7]. A tele-consultation session canshow thevarious documents available:image of a patientorphysician,radiographicimages,andotherclinicaldata.
Insomecircumstances,remotebiotechnologytoolscanbeusedto performenricheddiagnosticsbymoleculartechniques,targeting ofpathogens,andidentificationofsensitivitiestodrugsinorderto enhancetreatment[8].
Therestofthepaperisstructuredasfollows.Section2exposes thematerialandmethodswiththeconceptualgraphoperations usedtoimplementthemodellingofexpertrulesincollaborative decision-makingprocesses.Section3presentstheresultswithan illustrative case for the telemedicine management within the geriatricsfieldispresented.Finally,section4providesadiscussion andsection5concludesanddiscusseslessonslearnedandfuture challenges.
2. Materialandmethods
2.1. Knowledgeformalizationwithconceptualgraphs
Theconceptualgraph(CG)formalismisaknowledgerepresen- tationlanguage, which hasa well-definedsyntaxand a formal semantics that allows one to reason from its representations [9].Theconceptualgraphformalismisconsideredasacompromise representation between a formal language and a graphical language, because it is visual and has a range of reasoning processes[10].Conceptualgraphscanbeusedinmanycomputer science areas, including text analysis, web semantics, and intelligentsystems[11].
A simple conceptual graph is a finite, connected, directed, bipartitegraph consistingofconceptnodes (denotedas boxes), which are connected toconceptual relation nodes (denoted as circles). In the alternative linear notation, concept nodes are writtenwithin []-brackets while conceptualrelation nodes are denotedwithin()-brackets.Theconceptssetandtherelationsset aredisjoint.
A concept is composed of a type and a marker [<type>:
<marker>],forexample[Disease:IdiopathicParkinson’sdisease].
Thetypeofconceptrepresentstheoccurrenceofobjectclass.They aregroupedin a hierarchicalstructure calleda conceptslattice showing their inherit relationships. The marker specifies the meaningofaconceptbyspecifyinganoccurrenceofthetypeof concept.
Aconceptualrelationbindstwoormoreconceptsaccordingto thefollowingdiagram:
C1
½ ðrelation0snameÞ ½C2ðmeaningthat00C1 isrelatedtoC2
bythisspecificrelation00Þ
Intheanalysisoftelemedicinemanagement,themostcommon relationsaredependencyrelations,specifically,causal,condition- al,temporal,andBooleanconnectives,suchasAND,alternating-OR and exclusive-ORrelations. An exampleof conceptual graph is showninFig.1:amedicalactivityistheagentofatelemedicine serviceandits durationis influencedbythediagnosticcriteria.
The semantics of the core and extended CGs is defined by a formal mapping to and from a common abstract syntax and
model-theoreticfoundationforafamilyoflogic-basednotations (ISO/IEC24707)[12].
Aderivationisafinitesequenceoftheseelementaryoperations thathaveaformalsemanticsbasedonalogicalinterpretation.As aresult,themeaningofgraphoperationsisdeterminedinlightof the derivation to be applied, based on a logical interpretation whichgivesfulleffecttothevisualreasoning[13].Thederivation hasoneofthreeconceivablepropertiesonthelogicalrelationship betweenastartinggraphuandtheresultinggraphv[14]:
2.1.1. Equivalence
Copyandsimplifyareequivalencerules,whichgenerateagraph
vthatislogicallyequivalenttotheoriginal:theknowledgeofu isincludedinvandtheknowledgeofvisincludedinu(logical meaninguvandvu).
2.1.2. Specialization
Joinandrestrictarespecializationrules,whichgenerateagraph
v thatimplies theoriginal:v contains morespecificknowledge thanu(logicalmeaningvu).
2.1.3. Generalization
Detachandunrestrictaregeneralizationrules,whichgeneratea graph vthat is implied by theoriginal: v contains less precise knowledgethanv(logicalmeaninguv).
Ontological knowledge provides a formal description of the studied system [15] with associated experiences and lessons learned[16,17].
2.2. Exploitationofconceptualgraphsrepresentationinreasoning
AformalknowledgemodelledbyCGsinexperiencefeedback processescanbeaveryusefultoolforconveyingaccuratemeaning to a collaborative workenvironment involving domainexperts [18].Foragivenapplication,severalviewpointsofexpertisemay beengagedincombination.Forexample,someinvestigationsto improve the availability of a geriatric health care system can involveexpertknowledgeintele-expertiseandassociatedremote practices.
During theknowledge modelling phase of the telemedicine rules,theuseofCGpropertieswillhelptoenrichthetelemedicine knowledgebaseinordertoeasetheiraccess,sharingandreuseby themembersofthetelemedicinemanagementintheirindividual andcollectivetasks.Furthermore,experiencegainedandlessons learnedfrominitialproblemsolvingdevelopmentswillbeapplied as soon as possiblein similar situations in othercollaborative telemedicine activities with temporal modelling considerations [19].
In conceptualstructures, the operations of visual reasoning with semantic mapping form the bridge from perception to differentformsofconceptualoperations,rangingfromspecializa- tionandgeneralizationtounification[20].Whentheconceptual graphs operations are used in the visual reasoning, semantic comparisonsaredeployedateverystep,andtheonlydifference betweendeductionandanalogiesisthenatureoftheorientations onthereasoning[21].Particularly,oneconceptualgraphissimilar toanotherifthereisasemanticmapping(graphhomomorphism) fromthefirstgraphtothesecondone.Inthiscontext,thecase- basedreasoningcanbeengagedtosearchsomeunknownfeatures ofanewcasefromitsknownfeaturesandpreviouscasesstoredin thecasesbase.In analogicalreasoning,theconceptualstructure thatdescribesknownfeaturesofthenewcaseiscomparedwith thematchingfeaturesoftheconceptualstructuresassociatedto previous cases [22]. The assessment takes into account the determineddegreeofsimilarityandclassifiedalternativeoptions forothercases.Thecasethatprovidesthegreatestcorrespondence
maybeundertakenasthetopbasisofindicationforapproximating theunknownfeaturesofthenewcase.
3. Results
WeillustratetheproposedapproachusingCGswiththecase studyofsynucleinopathiesmostlyneurologicorgeriatricdiseases.
Theaccommodationoftheagingpopulationisamajorconcernin healthcareandforfamilies[23].Thecostofmedicalinterventions fortheelderly,whoareoftenfrailandwithrecurringepisodesof chronicdiseases,isincreasing[24].Supportforthesevulnerable people,whoareoftenisolated,requiresthedeploymentofflexible and effective systems to prevent hospitalizations sometimes experiencedbyelderlyasatraumaticeventandenabletheelderly to maintain their living place if possible [25]. Could new technologiesprovidesomesolutions?Theexpectedbenefitsare summarized by economic gain or improvement in the care of elderlypatientsortheirqualityoflife.Alsoarehealthcareteams theyreadytoembracethesenewtechnologies?
3.1. Clinicaloutlineofthesynucleinopathies
Thisapplicationisinthecontextofdecisionsupportinhealth andrelatesparticularlytoneuroscienceandespeciallysynuclei- nopathies,agroupofneurodegenerativediseasescharacterizedby fibrillary aggregates of alpha-synuclein protein (the major componentsofLewybodies,dystrophic(Lewy)neuritis,andthe Papp-Lantosfilaments)inthecytoplasmofselectivepopulationsof neurons and glia. These disorders include mainly Parkinson’s disease(PD),dementiawithLewybodies(DLB),multiplesystem
atrophy (MSA), and also pure autonomic failure (PAF) and Hallervorden-Spatz syndrome (HHS) [26]. Clinically, they are characterized by a chronic and progressive decline in motor (mainly Parkinsonism), cognitive, behavioral, and autonomic functions, dependingon thedistributionofthelesions[27].Al- though there are validated clinical and pathological consensus criteria for PD, DLB, and MSA, because of clinical overlap, differentialdiagnosisissometimesverydifficult[28].Moreover, early diagnosis of these diseases is also difficult [29]. Several approachesandnewbiomarkersarebeinginvestigatedtofacilitate the early ante-mortem diagnosis, as the common hallmark of synucleinopathies is the finding of alpha-synuclein protein on brain biopsy.Amongthesepotentialnewbiomarkers,there are myocardialmetaiodobenzyl guanidine(MIBG)scintigraphy[30], functional cerebral imageries (reduced dopamine transporter activity in some part of the brain), olfaction’s tests threshold revealinghyposmiaandolfactorypathwaypathology[31].Others newdiagnosiscorefeaturesasRapidEyeMovementsleep(REM sleep), behavior disorder, severe neuroleptic sensitivity, some clinical signs, symptoms and biological elements are assessing thestructureofthesefeaturesagainstthecriteriasetoutinthe guideline.However,therediagnosisaccuracyisnotoptimal,but inmostofcases,theycontributetoeliminatedifferentialdiagnosis.
Itwillbeusefultoinvestigatednewsapproachesadaptedtothese newneedswithpossibilitytoimprovediagnosiscriteria.
InParkinson’sdisease(PD)forexample,non-motorsymptoms are (described in Table 1): olfactory dysfunction, depression, constipation, pain,genitourinary problems,sleepdisorders, and cognitive impairments. These are often long ignored, and are common [32]. They sometimes precede motor symptoms and should be thoroughly investigated because they lead to a
Binary-Relation
Before During Temporal-relation
Usual-relation
Spatial-relation Logical-relation
Agent
Characterization Object
Implication Influence
Concept
Diagnosis Treatment Medical Activity
Diagnostic Criteria Sign Symptom Test result Telemedicine
Teleconsultation Teleexpertise Telemonitoring Teleassistance
Duration Competency Direction Tool Object
Telemedicine
Medical Activity agent Usual-relation Duration
Diagnostic Criteria influence Outside
Inside
Teleexpertise
Diagnosis: Idiopathic Parkinson'sdisease agent characterization Duration
Test Result influence Specialization Generalization
Patient
Fig.1.Examplesofconcepts/relationshierarchiesandconceptualgraph.
significantburden on the quality of life and have a significant contributiontothemorbidity ormortalityrates ofPD. Besides, functional imaging using dopaminergic tracers with either PositronEmissionTomography(PET)orSinglePositronEmission Tomography(SPECT)canidentifydopaminedeficienciesbutmay notreliablydifferentiatebetweenParkinson’sdiseaseandother akineticrigiddisorders.Inaddition,evenapathologicalexamina- tioncannotclassifyclinicalsyndromeswithcertainty.Despitethe diagnosticcriteria, a diagnosis of certainty is oftendifficult to establish because it requires a formal demonstration of the characteristicfeaturesofthepresumptivedisease.Clinicopatho- logicalstudieshaveshownthattheaccuracyofaclinicaldiagnosis ofPD isless than 80%.More selecteddiagnosticcriteria would increasetheproportionoftruePDcasesidentifiedtomorethan 90%[33].
Misdiagnosisofasynucleinopathyasanotheroneistherefore potentially extremely dangerous for these patients because of thetherapeuticsimplications.Forexample,inpatientswithPDor withdementia with Lewy bodies (DLB), typical neuroleptic or antipsychoticdrugscouldleadtoneurolepticmalignantsyndrome (NMS),alife-threateningneurologicaldisordermostoftendueto adversereactionsofthesedrugs.NMStypicallyconsistsofmuscle rigidity,fever,autonomicinstability,andcognitivechangessuch as delirium, and is associated with elevated plasma creatine phosphokinase.
Morerecentstudiesadvocatethatlanguage,especiallyseman- tics,maybeanareaofcognitioncapableofdistinguishingearly- onset synucleinopathy patients from healthy controls and discriminating between various types of synucleinopathies [34].Particularly,languagedisturbancesin dementiahavebeen factually attributed to the degradation of stored knowledge, whereasitisfrequentlyassumedthatlanguagedeficitsinstroke aphasiamirror modality-specificimpairment ofaccesstointact conceptual knowledge [35]. These deficits can be an acquired disorderoflanguagecomprehension,production,and/orsymbolic knowledge.In addition tomemory and language impairments,
assessment in other cognitive domains can also highlight the influenceof semanticknowledge,for instance,ondecoding the physico-chemical properties of an odorant [36]. Understanding thesemantics of cognitiveimpairmentsin synucleinopathiesis thusessentialforanumberofarguments.Identifyingsimilarities and differences between synucleinopathies is the first step in evolving tests that are differentially sensitive to the distinct conditions,thusprovidingappropriatescreeningteststosupport clinicaldiagnosis.Understandingthequalityofimpairmentsalso providessignificantinformationforpatients andcarers, helping physicians to elaborate better coping skills and care strategies.
Besides,comparingthepropertiesofcognitiveimpairmentbetween patient groups had often contributed to an improvement in a semanticanalysisoftheprocessunderlyingexecutivecontrolfor understandingthenormalorganizationofcognitivefunctions[33].
3.2. Visualmodellingofanalysisproceduresinsynucleinopathies
Wecanthink aboutthecontributionofartificialintelligence toolsasameansofanalysistosupportmedicaltestsfortheearly diagnosisanddifferentiationofthesepathologies.Particularlyin this context, operations of conceptualgraphscan be crucial to reasoningonknowledgewithvisualexplanations.Thisknowledge can provide evidence and analysis relevant to the activities of diagnosesanditisveryimportantformedicalresearchingeriatrics orneurosciences.Tothatend,weplantoincludetheengagement of reasoningtechniquesto determineby sequential association rules[37]themostsignificantrulestoassistintheestablishment of themost appropriate diagnosisand to eliminate differential diagnoses. Shared practice rules face the challenge of creating a solid foundation to support the cognitive development of collectiveintelligence.Theserulescanbeannotatedusingdomain ontologies andevaluated byotherusers,who mightconsidera form of suggestion using intelligent components based on the annotationsassociatedwithasynucleinopathy.Forexample,given acollaborative actiononasynucleinopathy,decisionsregarding Table1
Descriptionofnon-motorsymptomsassociatedwithIdiopathicParkinson’sdiseaseadaptedfrom[42].
Non-motorsymptomsassociatedtoIdiopathicParkinson’sdisease Supposedbraindamage Psychological,psychiatricand
cognitivesymptoms
Depression,anxiety,panicattacks,feelinghopeless Locuscoeruleus(norepinephrine),inferiorraphe nucleus(serotonin),amygdala
Apathy,anhedonia,inattention,vegetativesymptomsof depression
Uncertain Psychosis,inappropriatebehaviorandotheradversebehavioral
effectsdisorders,obsessive-compulsivedisorders,delirium
Limbiccortex,amygdala Parkinsonismdementia,confusion,hallucinations(iatrogenic
factor)
Temporallobe,hippocampus,locuscoeruleus, amygdala,nucleusbasalisofMeynert(acetylcholine) Sleepdisturbances Sleep-onsetinsomnia,trueinsomnia,nighttimehallucinations,
rapideyemovementsleepbehaviordisorder,daytime sleepiness,sleepattacks
Subcorticalnucleus(pedunculopontineand subcoerulusnucleus),hypothalamus Sleepfragmentation,sleepapnea(central(CSA),obstructive
(OSA),andcomplexormixedsleepapnea),restlesslegs syndrome(RLS)
Uncertain
Dysautonomicsymptoms Bladder-sphincterdysfunctions(urinaryfrequencyandurinary incontinence,dysuria),excessiveperspiration,abnormalsexual behavior(impotence,hypersexuality),xeropthalmia
Dorsalnucleusofvagusnerve
Orthostatichypotension,post-prandialhypotension Vagusnerve,stellateganglion(orcervicothoracic ganglionorinferiorcervicalganglion)
Gastrointestinalsymptoms Drooling,ageusia(tasteloss),swallowingdisorders,dysphagia, gastroesophagealreflux,vomiting,nausea,constipation,fecal incontinence
Dorsalnucleusofvagusnerve(adrenaline)
Sensorysymptoms Olfactorydisturbance Olfactorybulb
Pain,paresthesias Uncertain
Others Fatigue,seborrhea(skinandgreasyhair),excessive perspiration,weightgain,weightloss
Uncertain
![[PDF] Cours complet pour apprendre Ruby | Formation informatique](data:image/gif;base64,R0lGODlhAQABAIAAAP///wAAACH5BAEAAAAALAAAAAABAAEAAAICRAEAOw==)