数字化WACOM图形板数据集诊断帕金森病的研究(IJITCS-V9-N12-6)

I.J.InformationTechnologyandComputerScience,2017,12,45-51PublishedOnlineDecember2017inMECS()DOI:10.5815/ijitcs.2017.12.06Copyright©2017MECSI.J.InformationTechnologyandComputerScience,2017,12,45-51AStudyontheDiagnosisofParkinson’sDiseaseusingDigitizedWacomGraphicsTabletDatasetKemalAkyolDepartmentofComputerEngineering,KastamonuUniversity,37100,Kastamonu,TurkeyE-mail:[e-mail]:04September2017;Accepted:22September2017;Published:08December2017Abstract—ParkinsonDiseaseisaneurologicaldisorder,whichisoneofthemostpainful,dangerousandnon-curablediseases,whichoccursatolderages.TheStaticSpiralTest,DynamicSpiralTestandStabilityTestonCertainPointrecordswereusedintheapplicationwhichwasdevelopedforthediagnosisofthisdisease.Thesedatasetsweredividedinto80-20%trainingandtestingdatarespectivelywithintheframeworkof10-foldcrossvalidationtechnique.TrainingdataastheinputdataweresenttotheRandomForest,LogisticRegressionandArtificialNeuralNetworksclassifieralgorithms.Afterthisstep,performancesoftheseclassifieralgorithmswereevaluatedontestingdata.Also,newdataanalysiswascarriedout.Accordingtotheresultsobtained,ArtificialNeuralNetworksismoresuccessfulthanRandomForestandLogisticRegressionalgorithmsinanalysisofnewdata.IndexTerms—Parkinsondisease,digitizedWacomgraphicstablet,StaticSpiralTest,DynamicSpiralTest,StabilityTestonCertainPoint.I.INTRODUCTIONParkinson’sdisease(PD)isaneurologicaldisorder,whichisoneofthemostpainful,dangerousandnon-curablediseases,whichoccursatolderages.So,ithasnegativeimpactsonpatients’dailylife[1].Acommondiseaseofthecentralnervoussystemamongtheelderly[2],Parkinson’sdisease(PD)iscurrentlyincurable[3].Thecorrectdiagnosisofthisdiseaseisverydifficult[4],butpropertreatmentofthisdiseasecaneasethesymptomsandimprovethequalityoflifeofpatientssignificantly[3].Itscomplexsymptomsbringaboutchallengesfortheclinicaldiagnosis[2].Thus,itsefficientmonitoringandmanagementareveryimportant[3].Digitalanalysisofwritinganddrawingturnedintoavaluableresearchandclinicaltoolforthepatientswithessentialtremor,Parkinson’sdisease,dystonia,andrelateddisorders[5].Digitizedspiraldrawingcorrelateswithmotorscoresandmightbemoredelicateinearlychangesthansubjectiveratings[6].Spiraldrawingisusedinassessingoftremorseverityfortheclinicalevaluationofmedicaltreatments[7].Sakaretal.[8]collectedawiderangeofvoicesamplesforbuildingpredictivetelediagnosisandtelemonitoringmodels.TheyinvestigatedmorePD-discriminativeinformation,usingwell-knownmachinelearningtoolsandpublishedthedatasetforresearchers.Inthisstudy,theSST,DSTandSTCPdatasetswereobtainedfromthispubliclyrawdatasource.Thesedatasetsweredividedinto80-20%trainingandtestdatarespectivelywithintheframeworkof10-foldcrossvalidationtechnique.Thediagnosisofthisdiseasewasexploredbyutilizingthesedatasets.Next,bysendingthesedatasetsasinputdatatotheclassifieralgorithms,successesofthesealgorithmswereanalyzed.Therestofthispaperwasorganizedasfollows:InSection2,relatedstudieswereexamined.InSection3and4,informationaboutthemachinelearningalgorithmsandstatisticalanalysismetricswhichwereusedinthedesignedapplicationweregivenrespectively.InSection5,theinformationaboutthedatawasgivenindetail.InSection6,designprocessandresultswerepresented.Finally,theconclusionwasgiveninSection7.II.RELATEDWORKSTherearenumerousstudiesonthisdisease.Someofthestudiesareasfollows:In[2],theauthorsproposedanewmethodbasedonapolarcoordinatesystemwithvaryingorigininordertoquantitativelyevaluatetheperformanceinspiraldrawingtasksforpatientswithPD.Fifteensubjects,threeofwhomnormalsubjectsandtwelveofwhomarePDpatients,wereincludedinthedesignedspiraldrawingexperiment.Thehandmovementsofpatientswererecordedbyutilizingthedigitizedtabletrespectively.Thevariationoforigin,radius,degreeandothercharacteristicsofhandmovementswereevaluatedbyintroducingasetofparametersforfeatureextraction.Theresultshowedthattheproposedpolarcoordinatesystemembracedgoodperformanceinthequantitativeevaluationofspiraldrawing.In[3],theauthorsproposedamachinelearningmodelthatdetectsbradykinesiaanddyskinesiaforself-monitoringthemotorfunctioninParkinson'sdisease.Bradykinesia,whichistheslownessofmovement,istypicallyassociatedwithunder-medication.Dyskinesiawhichisinvoluntarymovementscanbetheresultofover-medication.Morespecifically,theaimwastoobjectivelyassessmotorsymptomsrelatedtobradykinesiasanddyskinesias.Thisworkcombinedspirographydataandclinicalassessmentsfroma46AStudyontheDiagnosisofParkinson’sDiseaseusingDigitizedWacomGraphicsTabletDatasetCopyright©2017MECSI.J.InformationTechnologyandComputerScience,2017,12,45-51longitudinalclinicalstudyinSwedenwiththefeaturesandpre-processingmethodologyofaSlovenianspirographyapplication.Overthecourseofthreeyears,over30.000spiral-drawingmeasurementswereinvolvedon65advancedPDpatients,and86%classificationaccuracyandover0.90AUCwereachieved.In[4],TheauthorsdevelopedanewdatasetbasedonhandwrittenexamstoaiddiagnosisofParkinson'sDisease.Thefeatureswereextractedfromthisdatasetovertwodifferentdrawings:spiralsandmeanders.Thisapp