Using Virtual Humans to Simulate Limited English Proficient Humans

MISSING IMAGE

Material Information

Title:
Using Virtual Humans to Simulate Limited English Proficient Humans
Physical Description:
1 online resource (79 p.)
Language:
english
Creator:
El Sheikh, Doaa T.
Publisher:
University of Florida
Place of Publication:
Gainesville, Fla.
Publication Date:

Thesis/Dissertation Information

Degree:
Master's ( M.S.)
Degree Grantor:
University of Florida
Degree Disciplines:
Computer Engineering, Computer and Information Science and Engineering
Committee Chair:
Lok, Benjamin C
Committee Members:
Peters, Jorg
Black, Erik W.

Subjects

Subjects / Keywords:
english -- human -- limited -- patient -- pediatric -- proficient -- virtual
Computer and Information Science and Engineering -- Dissertations, Academic -- UF
Genre:
Computer Engineering thesis, M.S.
Electronic Thesis or Dissertation
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )

Notes

Abstract:
Virtual human technology has been recently exploited in medical training. In this thesis, we propose using virtual humans to simulate limited English-proficient humans to assess the quality of health care provided to pediatric patients whose parents have limited English proficiency.  Through a user study, virtual humans were used to represent parents with different levels of English proficiency in a virtual pediatric interview setting. In the user study, medical students and residents interacted with virtual parents of different levels of English proficiency. The performance of medical students and residents when interviewing an English-proficient (EP) virtual mother and alimited English-proficient (LEP) mother was analyzed to assess the quality ofthe provided health care.  The results lead to some interesting conclusions. Firstly, virtual humans can effectively simulate different English proficiency levels. Secondly, the performance of the study participants regarding eliciting important information from patients was significantly higher with English-proficient patients. This conclusion agrees with the nationally observed fact that limited English-proficient patients getlower quality of health care than English-proficient patients mainly due to patient-provider language discordance. Thirdly, study participants learned howto recast their questions to adapt to the English proficiency level of the virtual mothers.
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (M.S.)--University of Florida, 2012.
Local:
Adviser: Lok, Benjamin C.
Statement of Responsibility:
by Doaa T. El Sheikh.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Classification:
lcc - LD1780 2012
System ID:
UFE0045124:00001


This item has the following downloads:


Full Text

PAGE 1

USINGVIRTUALHUMANSTOSIMULATELIMITEDENGLISHPROFICIENTHUMANSByDOAAT.ELSHEIKHATHESISPRESENTEDTOTHEGRADUATESCHOOLOFTHEUNIVERSITYOFFLORIDAINPARTIALFULFILLMENTOFTHEREQUIREMENTSFORTHEDEGREEOFMASTEROFSCIENCEUNIVERSITYOFFLORIDA2012

PAGE 2

c2012DoaaT.ElSheikh 2

PAGE 3

Idedicatethistomyparents,myhusbandandmykids,JumanahandAli. 3

PAGE 4

ACKNOWLEDGMENTS Thisthesiswouldnothavebeenpossiblewithouttheguidanceandthehelpofseveralmentorsandcolleagueswhocontributedandextendedtheirvaluableassistanceinthepreparationandcompletionofthiswork.Firstandforemost,myutmostgratitudetomyadviserDr.BenjaminC.Lok,whoseadviceandencouragementweretrulyinvaluable.Dr.RyanMoran,whodedicatedhistimeandefforttomakethisstudysuccessfulandhelpedwithalltheneededmedicalinformationgathering.Dr.ErikBlack,forhisinvaluablefeedbackandhisgenerousfundingforthestudy.IwouldliketothankDr.JorgPetersandDr.ErikBlackforacceptingtoserveinmycommitteeandprovidingtheirtimeandexpertisetoreviewmythesis.IwouldliketothankmycolleaguesattheVirtualExperiencesResearchGroup(VERG)fortheircontinuousassistanceandfeedback. 4

PAGE 5

TABLEOFCONTENTS page ACKNOWLEDGMENTS .................................. 4 LISTOFTABLES ...................................... 9 LISTOFFIGURES ..................................... 10 ABSTRACT ......................................... 12 CHAPTER 1VIRTUALHUMANSANDLIMITEDENGLISHPROFICIENTPATIENTS .... 13 1.1VirtualHumans ................................. 13 1.2LimitedEnglishProcientPatients ...................... 13 1.3TheProposedWork .............................. 14 2PREVIOUSWORK .................................. 16 2.1VirtualHumans ................................. 16 2.1.1VirtualHumansinTheMedicalField ................. 16 2.1.2VirtualHumansinMilitaryandSocialandCulturalTraining ..... 16 2.2LEPLatinosandDisparitiesinHealthCareQuality ............. 18 2.3ImprovingHealthCareQualityforLEPPatients ............... 18 3THEVIRTUALPEDIATRICSIMULATOR ...................... 19 3.1SystemOverview ................................ 19 3.2ScriptDatabase ................................ 20 3.2.1VPF ................................... 20 3.2.2CreatingtheHeartCaseScript .................... 21 3.2.3CreatingtheLungCaseScript .................... 21 3.2.4CreatingtheLEPScripts ........................ 22 3.3InterpersonalSimulatorandtheVirtualEnvironment ............ 23 3.3.1TheExamRoomandtheVirtualHumans .............. 23 3.3.2InterviewingtheVirtualMother .................... 25 3.3.3TheVirtualPhysicalExam ....................... 27 3.4TheUsedEquipment .............................. 28 4LANGUAGEBARRIERSSTUDY .......................... 30 4.1StudyDesign .................................. 30 4.2Hypotheses ................................... 30 4.3SecondaryExpectedFindings ........................ 31 4.4StudyProcedure ................................ 31 4.5Metrics ..................................... 32 4.5.1PercentageofIdentiedDiscoveries(ImportantFindings) ..... 32 5

PAGE 6

4.5.2TechniqueAdaptationtoRephrasingQuestionsforLEPCases .. 32 4.5.3DifferentialDiagnosisScore ...................... 33 4.5.4NumberofQuestionsAskedbyParticipants ............. 33 4.5.5SurveyResponses ........................... 34 5STUDYRESULTSANDDISCUSSION ....................... 36 5.1PrimaryHypothesis1andPercentageofIdentiedDiscoveries ...... 36 5.1.1ClassADiscoveries .......................... 36 5.1.1.1Three-wayinteractioneffect ................. 36 5.1.1.2Two-wayinteractioneffect .................. 36 5.1.1.3Maineffects ......................... 37 5.1.2ClassBDiscoveries .......................... 40 5.1.2.1Three-wayinteractioneffect ................. 40 5.1.2.2Two-wayinteractioneffect .................. 40 5.1.2.3Maineffects ......................... 42 5.1.3ClassCDiscoveries .......................... 43 5.1.3.1Three-wayinteractioneffect ................. 43 5.1.3.2Two-wayinteractioneffect .................. 43 5.1.3.3Maineffects ......................... 44 5.2PrimaryHypothesis2andPercentageofIdentiedDiscoveries ...... 45 5.2.1ClassADiscoveries .......................... 46 5.2.1.1Two-wayinteractioneffect .................. 46 5.2.1.2Maineffects ......................... 46 5.2.2ClassBDiscoveries .......................... 47 5.2.3ClassCDiscoveries .......................... 47 5.3PrimaryHypothesis3andTechniqueAdaptationtoRephrasingQuestions 47 5.4SecondaryFindings .............................. 48 5.4.1DifferentialDiagnosisScore ...................... 48 5.4.1.1Three-wayinteractioneffect ................. 48 5.4.1.2Two-wayinteractioneffect .................. 48 5.4.1.3Maineffects ......................... 50 5.4.2NumberofQuestions .......................... 52 5.4.2.1Three-wayinteractioneffect ................. 52 5.4.2.2Two-wayinteractioneffect .................. 52 5.4.2.3Maineffects ......................... 53 5.4.3IHavetheNecessarySkillsandKnowledgetoElicitaFocusedHistoryofThisPatient(Q1) ...................... 55 5.4.3.1Three-wayinteractioneffect ................. 55 5.4.3.2Two-wayinteractioneffect .................. 56 5.4.3.3Maineffects ......................... 56 5.4.4IKnewWhatPertinentInformationWasNeededtoMaketheDiagnosisforThisPatient(Q2) .......................... 57 5.4.4.1Three-wayinteractioneffect ................. 57 5.4.4.2Two-wayinteractioneffect .................. 57 6

PAGE 7

5.4.4.3Maineffects ......................... 58 5.4.5ItWasEasytoGettheInformationIWantedFromtheMother(Q3) 58 5.4.5.1Three-wayinteractioneffect ................. 58 5.4.5.2Two-wayinteractioneffect .................. 59 5.4.5.3Maineffects ......................... 59 5.4.6TheMotherGaveClearAnswerstoAllofMyQuestions(Q4) ... 60 5.4.6.1Three-wayinteractioneffect ................. 60 5.4.6.2Two-wayinteractioneffect .................. 60 5.4.6.3Maineffects ......................... 61 5.4.7IDidNotNeedtoRephraseorClarifyMyQuestionstotheMother(Q5) ................................... 62 5.4.7.1Three-wayinteractioneffect ................. 62 5.4.7.2Two-wayinteractioneffect .................. 62 5.4.7.3Maineffects ......................... 62 5.4.8IFeelCondentAbouttheDiagnosisBasedontheHistoryAlone(Q6) ................................... 63 5.4.8.1Three-wayinteractioneffect ................. 63 5.4.8.2Two-wayinteractioneffect .................. 63 5.4.8.3Maineffects ......................... 64 5.4.9ElicitingtheHistoryWasEasy(Q7) .................. 64 5.4.9.1Three-wayinteractioneffect ................. 64 5.4.9.2Two-wayinteractioneffect .................. 65 5.4.9.3Maineffects ......................... 65 5.4.10TheMotherAlwaysUnderstoodMyQuestions(Q8) ......... 67 5.4.10.1Three-wayinteractioneffect ................. 67 5.4.10.2Two-wayinteractioneffect .................. 67 5.4.10.3Maineffects ......................... 67 5.4.11TheMotherWasAlwaysAbletoAnswerMyQuestions(Q9) .... 67 5.4.11.1Three-wayinteractioneffect ................. 67 5.4.11.2Two-wayinteractioneffect .................. 67 5.4.11.3Maineffects ......................... 68 5.4.12INeverHadtoRephraseMyQuestionsinOrdertoGetInformationFromtheMother(Q10) ......................... 69 5.4.12.1Three-wayinteractioneffect ................. 69 5.4.12.2Two-wayinteractioneffect .................. 69 5.4.12.3Maineffects ......................... 70 6LIMITATIONANDFUTUREWORK ......................... 71 6.1Limitations ................................... 71 6.1.1LimitedDatabase ............................ 71 6.1.2DifcultyofMedicalCases ....................... 71 6.1.3LimitedNumberofParticipants .................... 71 6.2FutureWork ................................... 71 6.2.1ImprovingtheDatabase ........................ 71 7

PAGE 8

6.2.2HavingtheSystemOnline ....................... 72 6.2.3GivingImmediateFeedback ...................... 72 6.2.4PossibleStudies ............................ 72 6.2.4.1OrdereffectonLEPpatients ................ 72 6.2.4.2ImprovingthequalityofhealthcareforLEPpatients ... 73 7CONCLUSION .................................... 74 REFERENCES ....................................... 75 BIOGRAPHICALSKETCH ................................ 78 8

PAGE 9

LISTOFTABLES Table page 5-1PercentageofidentieddiscoveriesfromtherstandthesecondinterviewedLEPmothersforeachclassofdiscoveries ..................... 46 5-2Techniqueadaptationtorephrasingquestionsfortwoparticipants. ....... 48 9

PAGE 10

LISTOFFIGURES Figure page 1-1Thefourmedicalpediatriccases .......................... 15 3-1Systemoverview ................................... 19 3-2QuestionsandresponsesinaVPFscript ..................... 20 3-3Theheartpatientandhismother. .......................... 24 3-4Thelungpatientandhismother. .......................... 25 3-5Thispictureshowsthevirtualmother'sresponsetotheuser'squestion. .... 26 3-6Thispictureshowsthedisplayedspeechclaricationincasetheuser'squestionishaveyounoticedanychangeinskincolor?. .................. 26 3-7Fifteendifferentphysicalexamscanbeperformedonthevirtualpatient. .... 27 3-8AuserperformingtheHeartandLungsExamonthevirtualpatient ....... 28 3-9AuserperformingtheEarsExamonthevirtualpatient .............. 29 4-1Studyprocedureblockdiagram ........................... 32 5-1TheinteractioneffectbetweenEnglishprociencyandcaseorderforclassAdiscoveries. ...................................... 37 5-2ResultsforpercentageofidentiedclassAdiscoveriesfromtheEPandtheLEPmothers. ..................................... 38 5-3ResultsforpercentageofidentiedclassAdiscoveriesfromthemothersintheHeartandtheLangCases. ........................... 40 5-4TheinteractioneffectbetweenEnglishprociencyandcaseorderforclassBdiscoveries. ...................................... 41 5-5ResultsforpercentageofidentiedclassBdiscoveriesfromtheEPandtheLEPmothers. ..................................... 43 5-6ResultsforpercentageofidentiedclassBdiscoveriesfromthemothersintheHeartandtheLangCases. ........................... 44 5-7ResultsforpercentageofidentiedclassCdiscoveriesfromtheEPandtheLEPmothers,fromthemothersintheHeartandtheLangCasesandfromtherstandthesecondinterviewedmothers. ................... 45 5-8ResultsforpercentageofidentiedclassAdiscoveriesfromthemothersintheHeartandtheLangCases(forparticipantsinthesecondgroup). ...... 47 10

PAGE 11

5-9Theinteractioneffectbetweencasetypeandcaseorderforthedifferentialdiagnosisscore. ................................... 49 5-10ResultsofdifferentialdiagnosisscoreforEnglishprociency(EP,LEP). .... 51 5-11Histogramfordifferentialdiagnosisscores ..................... 52 5-12Resultsofdifferentialdiagnosisscoreforcasetype(Heart,Lung). ....... 53 5-13ResultsfornumbersofquestionsaskedtotheEPandtheLEPmothers. .... 54 5-14ResultsfornumbersofquestionsaskedtothemothersintheHeartandtheLungCases. ..................................... 55 5-15ResultsforQ1(Ihavethenecessaryskillsandknowledgetoelicitafocusedhistoryofthispatient). ................................ 57 5-16ResultsforQ2(Iknewwhatpertinentinformationwasneededtomakethediagnosisforthispatient). .............................. 58 5-17ResultsforQ3(ItwaseasytogettheinformationIwantedfromthemother). 60 5-18ResultsforQ4(Themothergaveclearanswerstoallofmyquestions). .... 61 5-19ResultsforQ5(Ididnotneedtorephraseorclarifymyquestionstothemother). 63 5-20ResultsforQ6(Ifeelcondentaboutthediagnosisbasedonthehistoryalone). 65 5-21ResultsforQ7(Elicitingthehistorywaseasy). .................. 66 5-22ResultsforQ8(Themotheralwaysunderstoodmyquestions). ......... 68 5-23ResultsforQ9(Themotherwasalwaysabletoanswermyquestions). ..... 69 5-24ResultsforQ10(Ineverhadtorephrasemyquestionsinordertogetinformationfromthemother). ................................... 70 11

PAGE 12

AbstractofThesisPresentedtotheGraduateSchooloftheUniversityofFloridainPartialFulllmentoftheRequirementsfortheDegreeofMasterofScienceUSINGVIRTUALHUMANSTOSIMULATELIMITEDENGLISHPROFICIENTHUMANSByDoaaT.ElSheikhDecember2012Chair:BenjaminLokMajor:ComputerEngineeringVirtualhumantechnologyhasbeenrecentlyexploitedinmedicaltraining.Inthisthesis,weproposeusingvirtualhumanstosimulatelimitedEnglish-procienthumanstoassessthequalityofhealthcareprovidedtopediatricpatientswhoseparentshavelimitedEnglishprociency.Throughauserstudy,virtualhumanswereusedtorepresentparentswithdifferentlevelsofEnglishprociencyinavirtualpediatricinterviewsetting.Intheuserstudy,medicalstudentsandresidentsinteractedwithvirtualparentsofdifferentlevelsofEnglishprociency.TheperformanceofmedicalstudentsandresidentswheninterviewinganEnglish-procient(EP)virtualmotherandalimitedEnglish-procient(LEP)motherwasanalyzedtoassessthequalityoftheprovidedhealthcare.Theresultsleadtosomeinterestingconclusions.Firstly,virtualhumanscaneffectivelysimulatedifferentEnglishprociencylevels.Secondly,theperformanceofthestudyparticipantsregardingelicitingimportantinformationfrompatientswassignicantlyhigherwithEnglish-procientpatients.ThisconclusionagreeswiththenationallyobservedfactthatlimitedEnglish-procientpatientsgetlowerqualityofhealthcarethanEnglish-procientpatientsmainlyduetopatient-providerlanguagediscordance.Thirdly,studyparticipantslearnedhowtorecasttheirquestionstoadapttotheEnglishprociencylevelofthevirtualmothers. 12

PAGE 13

CHAPTER1VIRTUALHUMANSANDLIMITEDENGLISHPROFICIENTPATIENTS 1.1VirtualHumansVirtualhumansareconversationalembodiedagentsthatlookandactlikehumans[ 16 ].Virtualhumanshavebeenbuilttobeusedinavarietyofapplications.Virtualhumanshavebeenusedindifferenttypesofinterpersonalskillstrainingsuchasinthemilitary[ 9 20 ],negotiation[ 8 34 ],medicaleducation[ 19 24 ],publicspeaking[ 28 ]andteamtraining[ 30 ].Virtualhumanshavebeenusedasvirtualpatients.Virtualpatientsarevirtualhumansthatplaytheroleofstandardizedpatients.Standardizedpatientsareactorsthatplaytheroleofpatientsfortrainingmedicalstudents.Virtualhumanshavebenetsoverstandardizedpatientsasvirtualpatientscansimulateconditionsthatstandardizedpatientsareunabletosimulate.Forexample,virtualpatientscansimulateapatientwithcranialnervepalsywhichisdifcultforstandardizedpatientstosimulate[ 3 ].Virtualhumanscansimulatehumanswithdifferentracestoelicitracialbias[ 31 ],simulatehumanswithdifferenttypesofaudiencebehaviorstoelicitanxiety[ 28 ]andsimulatehumanswithdifferentlevelsofEnglishprociencyaswewillproposeinthisthesis. 1.2LimitedEnglishProcientPatientsIn2010,25.2millionindividualsintheU.S.,amountingroughlytoninepercentofthecountry'spopulation,wereconsideredtobeoflimitedEnglishprociency[ 26 ].LimitedEnglish-procient(LEP)patientsarelesslikelythanEnglish-procient(EP)patientstoobtainsufcientinformationfromhealthcareprovidersandtobeencouragedtoparticipateinmedicaldecisionmaking[ 13 ].Asaresult,LEPpatientstypicallyreceivelowerqualityofhealthcaredespitetheirlevelofincomeoraccesstohealthcareinsurance[ 7 ]. 13

PAGE 14

Assessmentofthequalityofhealthcareprovidedtopatientsisusuallyconductedthroughsurveyslledbypatientsinclinicsorbyphone[ 10 29 ].Asaresult,assessmentofthequalityofhealthcareprovidedtopatientsdependsonlyonthepatients'opinionandignoresthephysicians'sideandhencetheassessmentmaynotbeaccurate.ThelevelofEnglishprociencyinapatientisusuallyassessedusingthepatient'sanswertothequestionHowwelldoyouspeakEnglish?.Ifthepatientanswerswithverywellorwell,thepatientisconsideredtobeanEP.Incasethepatientanswerswithnotwellornotatall,thepatientisconsideredtobeanLEP[ 1 ]. 1.3TheProposedWorkThisthesisproposestheuseoflimitedEnglishprocient(LEP)virtualhumanstoaddresstheimpactoflanguagebarriersontheperformanceofhealthcareproviders.Toaddresstheimpactoflanguagebarriersontheperformanceofhealthcareproviders,LEPvirtualhumanswereusedtoassessandimprovethequalityofhealthcareprovidedtoLEPpatientsthroughassessingandimprovingthecommunicationskillsofphysicians.Fourpediatriccaseswerecreatedtohelpevaluateandimprovethecommunicationskillsofbothmedicalstudentsandpediatricresidents.ThefourpediatriccasesrepresentthecombinationsoftwomedicalcasesandtwolevelsofparentalEnglishprociency(Figure 1-1 ).Ineachofthefourcases,twovirtualhumanswereused:onetosimulatethechildpatientandanothertooktheroleofhismother.Inauserstudy,thirtyoneparticipantscomposedofmedicalstudentsandpediatricresidentsfromtheUniversityofFloridamedicalschoolwererecruited.Participantinteractedwithtwovirtualpediatricpatientsofdifferentmedicalcases(HeartandLungcases)andwithdifferentparentalEnglishprociencylevels(EPorLEP).TheresultsshowedthattherewasasignicantdifferencebetweentheperformanceofparticipantsregardingtheEnglishprociencylevelofthetwointerviewedmothers.TheresultsshowedthatparticipantswereabletoelicitmoreinformationfromtheEPmothersthantheLEPmothers. 14

PAGE 15

Figure1-1. Thefourmedicalpediatriccases Moreover,weexplainhowweusedvirtualhumanstoimprovecommunicationskillsoftheparticipantsforinterviewingLEPpatients.ParticipantsavoidedusingmedicalterminologyandusedsimplesentencesandcommonwordstoelicitinformationfromtheLEPparents.Participantsrephrasedtheirquestionswhenevertheparentdidnotunderstand. 15

PAGE 16

CHAPTER2PREVIOUSWORKInthefollowingtwosections,wepresentsomeofthepreviousapproachesthatrelatetoourideasof Usingvirtualhumans EvaluatingandimprovingthequalityofhealthcareforLEPpatients. 2.1VirtualHumans 2.1.1VirtualHumansinTheMedicalFieldJohnsenetal.[ 19 ]comparedtheperformanceofmedicalstudentswheninterviewingastandardizedpatientandwheninterviewingavirtualpatienttocheckforcorrelation.Acorrelationwasfoundbetweentheperformanceofstudentsinbothinterviews.Thiscorrelationindicatesthatvirtual-humaninterviewscanbeaseffectiveasreal-humaninterviewsandhencecanbeusedineducationtoimprovetheinterpersonalskillsofmedicalstudents.Parsonsetal.[ 27 ]createdasystemwithvirtualhumanssufferingfromamentaldisorder.Inthatsystem,healthcareprovidersweretrainedoninterviewingapatientwithamentaldisorderusingvirtualhumans.Thehealthcareproviderswhousedthesystemestablishedrapportwiththevirtualhumans.Theprovidershadusefuldiscussionswiththevirtualhumansabouttheirtraumaticexperiences.Rossenetal.[ 31 ]usedvirtualhumanswithdifferentskin-tonestoinvestigateskin-tonebias.VirtualhumanswereinterviewedbyCaucasianmedicalstudents.Eachmedicalstudentinterviewedeitherablackorawhitevirtualhuman.Virtualhumansshowedtheirabilitytoelicitbehaviorthatisconsistentwithreal-worldskin-tonebiases.Virtualhumanswereshowntobeagoodsupplementforanti-biastrainingmodules. 2.1.2VirtualHumansinMilitaryandSocialandCulturalTrainingIntheMissionRehearsalExercise(MRE)project,USC'sICTgroupusedvirtualhumanstoimprovethedecision-makingskillsofunitleadersintheU.S.Army[ 20 ]. 16

PAGE 17

IntheMREproject,participantsweretrainedonmakingcorrectdecisionsinsomesituationsbygivingthemtheopportunitytointeractwithalargerthanlife-sizevirtualhumans.TheMREprojecthelpsparticipanttobecomewellpreparedforsimilarsituationsinreallife.TheVirtualEnvironmentCulturalTrainingforOperationalReadiness(VECTOR)projectusesvirtualhumansforculturaltrainingintheU.S.military[ 9 ].TheVECTORprojecthelpsforcestodevelopspecicskillsneededforinteractingwithmembersofacultureofinterestandhencehelpstheseforcestosurviveinaforeignenvironment.Bickmoreelal.[ 6 ]usedananimatedcharacterasarelationalagent.Therelationalagentwasdesignedtotaketheroleofanexerciseadviserthatinteractwithelderlyusersdailyfortwomonths.Therelationalagentgoalwastomotivateuserstoexercisethroughwalking.Therelationalagentsystemprovedtobeeffectiveforhealtheducationapplicationsforelderlyuserswithlowcomputerliteracy.Babuetal.[ 4 ]usedvirtualhumansinastudytoinvestigatetheirpotentialtoteachpeopleIndiansocialconversationaltraditions.Virtualhumanstaughtparticipantsthesocialtraditionssignicantlybetterthantheliterature-basedmethods.Kangelal.[ 21 ]conductedastudyonrealhumanstoelicitthedifferenttypesofhumannonverbalcuesthatareassociatedwithdifferentintimacylevelsofverbalself-disclosure.Theelicitednonverbalcuesassociatedwithverbalself-disclosurewerethenusedtodesignavirtualcounselor.Thevirtualcounselorwiththenonverbalcueswasusedinanonlinesurveyinordertoevaluatetheeffectofthosenonverbalcues.Thevirtualcounselorwithappropriatenonverbalcuesforintimateself-disclosureprovedtobelikedbytheirhumanclients.Thenonverbalcuesindeedhelpedincreatingabetterrapporttowardsthosecounselors.Pertaubetal.[ 28 ]assessedtheanxietyresponsesofpeoplegivingpresentationstovirtualhumans.Thevirtualhumanshadthreedifferenttypesofaudiencebehaviors: 17

PAGE 18

neutral,positiveandnegative.Thevirtualhumanswereshowntobecapableofprovokingappropriatearousalinpeoplewithpublicspeakinganxiety. 2.2LEPLatinosandDisparitiesinHealthCareQualityTheunfavorableimpactofthelimitedEnglishprociencyisparticularlymoreprominentinLatinogroups.Decamp[ 10 ]usedthe2007NationalSurveyofChildrenHealth(NSCH)asthedatasourceforastudyaboutmedicalhomedisparitiesforLatinochildren.LatinochildrenwithaSpanish-languageparentalinterviewwerelesslikelytogetprimary-careaccessandgoodqualityofhealthcarethanLatinochildrenwithanEnglish-languageparentalinterview.Moreover,bothgroupsofLatinochildrenwerelesslikelythanWhitechildrentogetprimary-careaccessandgoodqualityofhealthcare.Pippinsetal.[ 29 ]showedthatinsuredLEPLatinosweremorelikelythaninsuredEPLatinostoreportnothavingaregularsourceofcare,lackingcontinuity,longwaitsanddifcultygettinginformationbyphone. 2.3ImprovingHealthCareQualityforLEPPatientsIncreasingaccesstolanguage-concordantphysicianscanimprovetheexperienceofLEPpatients[ 35 ].LEPpatientswereshowninpreviousstudiestobemorelikelytoreportbetterhealthcarewhenhavinglanguage-concordantphysicians[ 14 25 33 ].However,restrictingLEPpatientstolanguage-concordantphysiciansmayleadtolowerqualityofhealthcare.Thelowerqualityofhealthcareisduetothepossibilitythatlanguage-concordantphysiciansarelessqualiedandhavelessaccesstoimportantmedicalresources[ 5 ].ProfessionalinterpreterscanincreasethequalityofhealthcaredeliveredtoLEPpatients[ 15 17 22 ].However,interpretersmaybeunderusedevenwhenavailablebecausephysiciansprefertousetheirownlimitedsecond-languageskillsorfamilymembersasinterpreters[ 11 12 ].Moreover,evenwiththeuseofinterpreters,themajorityofcliniciansreporteddifcultiesdealingwithLEPpatients[ 23 ].Inaddition,LEP 18

PAGE 19

patientsmaybelesssatisedevenwiththeuseofinterpreterscomparedtopatientsthatareabletocommunicatedirectlywiththeirphysicians[ 25 ]. 19

PAGE 20

CHAPTER3THEVIRTUALPEDIATRICSIMULATOR 3.1SystemOverviewThevirtualpediatricsimulatoriscomposedoftwomaincomponents(Figure 3-1 ): Thescriptdatabasethatcontrolsthevirtualhumanresponsestotheuser'squestions. Theinterpersonalsimulator(IPS)[ 18 ]thatrendersthevirtualenvironment,connectstothescriptdatabaseandreceivestheuser'squestionstothevirtualhumanthroughagraphicaluserinterface. Figure3-1. Systemoverview 20

PAGE 21

3.2ScriptDatabase 3.2.1VPFPediatricinterviewingscriptsforthevirtualpediatricsimulatorwerecreatedusingVirtualPeopleFactory(VPF)[ 32 ].VPFisaweb-basedapplicationthatallowsdeveloperstocreatevirtualhumanscripts.Avirtualhumanscriptisacorpusofallpossiblequestionsthatausercanasktothevirtualhumanandtheirresponses.Eachresponseisconnectedtoagroupofquestions(Figure 3-2 ).Aresponseistriggeredwhenoneofitsconnectedquestions(orasimilarone)isaskedbytheuser.ThetriggeredresponseisdisplayedonthescreenintheuserinterfacepageofVPFandmayalsobeheardifthereisavoicerecordedforthatresponse. Figure3-2. QuestionsandresponsesinaVPFscript 21

PAGE 22

3.2.2CreatingtheHeartCaseScriptTherststepincreatingtheheartscriptwastochooseaspecicheartcase.Theheartcase(aswellasothercases)waschosenbyDr.RyanMoran,anassistantprofessorintheDivisionofPediatricCriticalCareattheUniversityofFlorida.Dr.Moranwroteadetaileddescriptionaboutthemedicalcase.Themedicalcasewasabouta12-month-oldboynamedBobby.BobbyissufferingfromMyocarditiswhichisinammationoftheheartmuscle.BobbyispresentintheclinicwithhismotherLinda.Bobby'scasewillbereferredtoastheHeartCase.Followingthecasedescription,35responsesand52questionswereaddedtoVPFtocreateascriptforLinda.ThismeansthatLindawasabletogive35differentresponsesthatweretriggeredbythe52addedquestions.ThesecondstepwastomakesomeinitialtestsonthecreatedscriptbyaskingquestionstoLinda.WheneverLindawasnotabletorespondtoaquestion,thequestionwasaddedtoasuitableresponseinthescriptdatabase.Iftherewasnosuitableresponseinthescriptdatabase,thequestionwouldbeaddedandconnectedtoanewsuitableresponse.Aftertheinitialtests,thecreatedscriptgrewtoinclude75responsesto435differentquestionsandLinda'sresponseswererecordedbyanEPfemale.Thelaststepwastotestthescriptbydomainexpertsandgeneralpeople.Asaresultofthisstep,newquestionsandresponseswereaddedtoLinda'sscript.Linda'sscriptreachedastablestatewheretherewasalmostnonewaddedresponseswithnewtesting.Thetestingwaspausedatthispoint.Currently,Linda'sscripthas216responsestodifferent1193questions. 3.2.3CreatingtheLungCaseScriptAsecondmedicalcasewaschosentobeofthesamelevelofdifcultyoftheHeartCase.Thesecondmedicalcaseisaboutan11-month-oldboynamedAlex.Alexhasswallowedaforeignbody(abigpieceofcarrot).Alexhassomecommonsymptomswith 22

PAGE 23

Bobbysuchashardbreathing.AlexispresentintheclinicwithhismotherTrisha.Alex'scasewillbereferredtoastheLungCase.TheHeartandtheLungCaseswerediscussedbetweenDr.Moranandsomecolleaguesinordertotestifthetwocasesareofthesamedifculty.TheHeartandtheLungCaseswereclassiedtobeofthesamedifculty.ThescriptfortheHeartCasewasreusedtocreateanewscriptforAlex'smother,Trisha.ThenewscriptwasmodiedtomatchtheLungCaseandTrisha'svoicewasrecordedbyanEPfemale(notthesamefemalethatrecordedLinda'svoiceintheHeartCase).ThenewscriptwascreatedfasterandwithlesstestingduetotheadvantageofreusingthescriptfortheHeartCase.Currently,Trisha'sscripthas222responsestodifferent1198questions.Thescriptscreatedforthetwomedicalcaseshasmostlythesamequestionsbutwithdifferentresponsesdependingoneachcase.Alsodependingoneachcase,therearesomedifferentquestions.ForexampleintheLungCasescript,thereisaquestionaboutthecoloroftheproducedmucous.ThisquestionisnotintheHeartCasescriptbecausetheheartpatientdoesnothavemucoussoitdoesnotmakesensetoaskaboutthecoloroftheproducedmucous. 3.2.4CreatingtheLEPScriptsTheHeartCasedescriptionwastranslatedintoSpanishandwasgiventoaSpanish-procientfemalewhoisanLEPperson.TheLEPfemalestudiedthetranslatedcasedescriptionandthenwasaskedallthequestionsintheHeartCasescriptinEnglish.TheLEPfemaleresponseswererecordedandusedastheresponsesforavirtualLEPmother(Maria)inanewscript.ThenewscriptisaboutCarloswhohasthesameexactsymptomsandconditionsasBobby(theheartpatient)butwithdifferentparents.TheLungCasedescriptionwastranslatedintoSpanishandwasgiventoadifferentLEPSpanish-procientfemale.TheLEPfemalestudiedthetranslatedcasedescription 23

PAGE 24

andthenwasaskedallthequestionsintheLungCasescriptinEnglish.TheLEPfemaleresponseswererecordedandusedastheresponsesforavirtualLEPmother(Marta)inanewscript.ThenewscriptisaboutAlejandrowhohasthesameexactsymptomsandconditionsasAlex(thelungpatient)butwithdifferentparents.ThosetwocreatedLEPscriptshavethesamequestionsastheirEPversionsbutwithdifferentresponsesduetothedifferenceintheEnglishprociencylevelsoftheLEPandtheEPmothers.Insomesituations,iftheinterviewerasksthemotherintheEPscriptacoupleofdifferentquestionsthathavethesamemeaning,he/shewillgetthesameexactresponse.Forexample,themother'sresponsetothequestionsAnyeardischarge?andAnyuidcomingfromhisear?willbethesame:Noeardischarge.However,theresponseofthemotherintheLEPscriptwillbeDischarge?Idon'tknowwhatthatmeansandNo,nothingiscomingfromhisear,respectively.ThevariationinresponsesfortheLEPscriptsteacheshealthcareprovidersthattheyneedtochoosetheirwordscarefully.HealthcareprovidersshouldalwaysavoidusinghardmedicalterminologywithLEPpatients. 3.3InterpersonalSimulatorandtheVirtualEnvironmentInterpersonalSimulator(IPS)isasoftwaresystemthatenablesuserstopracticecommunicationskillswithvirtualhumans[ 18 ].IPSwasusedtorenderfourvirtualenvironments.EachvirtualenvironmentisconnectedtoadifferentscriptdatabaseonVPFtocreatetwodifferentlevelsofparentalEnglishprociencyfortwodifferentmedicalcaseswhicharetheHeartandtheLungCases.Eachvirtualenvironmentrepresentsavirtualexamroomwithvirtualobjectsandtwovirtualhumansthatrepresentthepatientandthemother. 3.3.1TheExamRoomandtheVirtualHumansInthevirtualexamroom,thereexistsananimatedvirtualmother.Thevirtualmothercanrespondverballytotheuser'squestions.Thevirtualmotheraccompanies 24

PAGE 25

thepatientwhoisrepresentedbyananimatedvirtualbaby.Thevirtualbabysuffersfromaproblemthatcauseshimtohaveafastandheavybreathing.ThebreathingwascreatedinMayaandsimulatedbyIPS.ThemothersintheHeartandtheLungCaseswererepresentedbytwodifferentlookingvirtualfemalesthatcanbeperceivedaseitherCaucasianorLatinotoavoidracebias.ThemotherswereshowntotendifferentpersonswhoagreedthatthetwomotherscanbeconsideredeitherCaucasianorLatino.OnevirtualbabywasusedtoplaytheroleofthepatientintheHeartandtheLungCases.InboththeHeartandtheLungCases,thevirtualbabyhasamatchingeye,hairandskincolorswithhisvirtualmother.WithineachoftheHeartandtheLungCases,thesamevirtualfemalewasusedtorepresentboththeLEPandtheEPvirtualmothers(Figure 3-3 andFigure 3-4 ). Figure3-3. Theheartpatientandhismother. 25

PAGE 26

Figure3-4. Thelungpatientandhismother. 3.3.2InterviewingtheVirtualMotherByhavingthefourscripts,userscaninterviewthevirtualmothersthroughthegraphicaluserinterfaceofIPSandelicitinformationbyaskingrelevantquestions.ThesequestionsshouldberelatedtothehealthstatusofthepatientsuchasHashebeencoughing?andDoeshehaveanywheezing?.ThesequestionsalsoincludefamilyhistoryandsomegeneralinformationsuchasAnymedicalproblemsinthefamily?andDoyouhaveanyotherchildren?.Whenausertypesaquestiontothemotherthattriggersaresponseinthedatabase,themotherrespondsverbally.Inadditiontotheverbalresponseofthemother,atextofherresponseisdisplayedonthescreen(Figure 3-5 ).Incasetheuser'squestiontriggersdifferentresponses,aspeechclaricationisdisplayed.Thespeechclaricationdisplaysamaximumofthreedifferentpossiblequestionsthatthevirtualhumanthoughttheusermighthaveasked.Thespeech 26

PAGE 27

Figure3-5. Thispictureshowsthevirtualmother'sresponsetotheuser'squestion. Figure3-6. Thispictureshowsthedisplayedspeechclaricationincasetheuser'squestionishaveyounoticedanychangeinskincolor?. 27

PAGE 28

claricationalsogivestheusertheoptiontochoosethathe/shedidnotmeananyofthedisplayedquestionsortochoosethathe/shewastypingastatement(Figure 3-6 ).Whentheusergetsaresponsethathasanimportantpieceofinformation,adiscoveryisloggedinale.Forexample,ifthemother'sresponseisHeseemsreallycongestedinhischest,thentheloggeddiscoveryshouldbeBreathingischaracterizedwithchestcongestion. 3.3.3TheVirtualPhysicalExamInadditiontointerviewingthevirtualmother,theusercanperformfteendifferentphysicalexamsonthevirtualbaby(Figure 3-7 ). Figure3-7. Fifteendifferentphysicalexamscanbeperformedonthevirtualpatient. Thephysicalexamsaredividedintointeractiveandnon-interactiveexams: OneinteractiveexamwhichistheAuscultateHeartandLungsExam.Inthisexam,theusercanusethemousetomoveavirtualstethoscopeandlistentothebaby'sheartandlungsbylift-clicking(Figure 3-8 ).Dependingonwhatlocationtheuser 28

PAGE 29

listensto,he/shewilldetectdifferentndings.Thelocationofthestethoscopeisloggedperiodicallymanytimespersecondtohelpinevaluatingtheuser'sperformance. Fourteennon-interactiveexamswhicharetheGeneralExam,theHeadExam,theEarsExam,theEyesExam,theNoseExam,theThroatExam,theNeckExam,theAbdomenExam,theGenitourinaryExam,theBackExam,theAnalExam,theExtremitiesExam,theSkinExamandtheNeuroExam.Allthesefourteennon-interactiveexamsgiveinformationaboutwhattheuserwouldndincasehe/sheperformedtheexamonthevirtualbaby.Amongthoseexams,theEarsandtheThroatexamshavepicturesalongwiththeinformation(Figure 3-9 ). Figure3-8. AuserperformingtheHeartandLungsExamonthevirtualpatient 3.4TheUsedEquipmentTointeractwiththevirtualhumans,commodityover-the-shelfequipmentwereused: AMacMinicomputertorunIPSandrenderthevirtualenvironmentandthevirtualhumans. A15-inchstandardmonitorfordisplayingthevirtualenvironmentandthevirtualhumans. 29

PAGE 30

Figure3-9. AuserperformingtheEarsExamonthevirtualpatient Speakersforthevoiceofthevirtualmothersandtheheartandlungssounds. Akeyboardfortypingquestionstothevirtualmothers. AmouseforperformingtheHeartandLungsExamonthevirtualpatient. 30

PAGE 31

CHAPTER4LANGUAGEBARRIERSSTUDY 4.1StudyDesignAmixeddesignstudywasconductedonagroupofthirtyoneparticipants(elevenmalesandtwentyfemales)toevaluatethecommunicationskillsofparticipantswheninterviewingEPandLEPpatientsandtocheckifparticipantsdevelopedbettercommunicationskillswiththeLEPpatientsovertime.Twoparticipantswerethird-yearpediatricresidents,tenweresecond-yearpediatricresidents,threewererst-yearpediatricresidentsandsixteenwerethird-yearmedicalstudents.AllparticipantswererecruitedfromUniversityofFloridamedicalschool.Eachparticipantreceivedacompensationofa$10giftcardforparticipatinginthestudy.Participantswererandomlyplacedintotwodifferentgroups.Participantsineachgroupinteractedwithoneheartcaseandonelungcaseandtheorderofthecaseswaschosenrandomly.Intherstgroup,oneofthecaseswaschosenrandomlytohaveanLEPmotherandtheotheranEPmotherwhereasinthesecondgroupbothcaseswerechosentobeofLEPmothers.Ineachcase,participantsgothroughaninterviewsessionfollowedbyaphysicalexamsession.Afternishingthephysicalexam,participantsareallowedtogobacktotheinterviewandaskmorequestionstothevirtualmother. 4.2Hypotheses PrimaryHypothesis1:Participantswillgathermorecriticalndings(discoveries)fromthevirtualEPmothersthanthevirtualLEPmothers. PrimaryHypothesis2:ParticipantswillimprovetheircommunicationskillsfordealingwiththeLEPmothersafterusingthesystem. PrimaryHypothesis3:ParticipantswillidentifymoretechniquesfordealingwiththeLEPmothersafterusingthesystem. 31

PAGE 32

4.3SecondaryExpectedFindings ParticipantswillbemorelikelytocometothecorrectdiagnosisafterinteractingwiththevirtualEPmothersmorethantheywouldafterinteractingwiththeLEPmothers. ParticipantswillaskmorequestionstotheLEPmothersthantheEPmothers. Therewillbeadifferenceintheparticipants'responsestothesurveyquestionsbetweentheEPandtheLEPmothers. 4.4StudyProcedureThestudywasconductedinaconferenceroomatShandshospital.Eachparticipantwentthroughthefollowingsteps: Wasrandomlyassignedtoagroupandgivenaconsentformtoreadandsign. Watchedaninteractivetutorialwithavirtualdoctorexplaininghowtousethesystemandhowtoperformthephysicalexam.Theparticipanthadtheopportunitytopracticeinteractingwiththevirtualdoctorandperformtwoexamsonavirtualbaby. Filledapre-survey. Interviewedthevirtualmotherintherstcasetoelicitinformation. Filledinaninterimsurveyabouttherstcase. Performedthephysicalexamonthevirtualbabyintherstcase. Filledinanotherinterimsurveyabouttherstcase. Interviewedthevirtualmotherinthesecondcasetoelicitinformation. Filledinaninterimsurveyaboutthesecondcase. Performedthephysicalexamonthevirtualbabyinthesecondcase. Filledinanotherinterimsurveyaboutthesecondcase. Filledinapostsurvey. 32

PAGE 33

Figure4-1. Studyprocedureblockdiagram 4.5Metrics 4.5.1PercentageofIdentiedDiscoveries(ImportantFindings)ThepercentageofidentieddiscoveriesisusedfortestingPrimaryHypothesis1andPrimaryHypothesis2.Discoveriesaretheusefulinformationthataparticipantelicitsfromthevirtualmother'sresponses.Discoveriesweredividedbyamedicalexpert(Dr.Moran)intothreeclasseswereeachclasswasanalyzedseparately: ClassA:Theimportantinformationthatparticipantshavetoelicitfromthemothersandwhichhelpsinarrivingatthecorrectdiagnosis. ClassB:Theinformationthatparticipantsmayormaynotelicitfromthemothersandwhichhelpsinarrivingatthecorrectdiagnosis. ClassC:Theinformationthatparticipantsmayormaynotelicitfromthemothersandwhichdoesnothelpinarrivingatthecorrectdiagnosisbutitiselicitedineachpediatricinterview. 4.5.2TechniqueAdaptationtoRephrasingQuestionsforLEPCasesParticipant'susageofrephrasingisusedfortestingPrimaryHypothesis3. 33

PAGE 34

AnticipatedTechniqueScore:Inthepre-survey,eachparticipantwasaskedtoprovidethetechniqueshe/sheanticipatedtousetoelicittheinformationhe/sheneedsfromanLEPpatient.Iftheparticipant'sanswerincludesrephrasingorsomethingsimilarhe/shegetsapoint. UsedTechniqueScore:Inthepostsurvey,eachparticipantwasaskedtoprovidethetechniqueshe/sheindeedusedtoelicittheinformationhe/sheneededfromtheLEPvirtualmother.Iftheparticipant'sanswerincludesrephrasingorsomethingsimilarhe/shegetsapoint. TechniqueAdaptationScore:Thedifferencebetweentheusedtechniquescoreandtheanticipatedtechniquescoregivesthetechniqueadaptationscore.Ifthetechniqueadaptationscoreispositive,thentheparticipantlearnedfromthesystemhowtorephrasequestionsforLEPpatients.Otherwise,norephrasingtechniquewaslearnedfromthesystem. 4.5.3DifferentialDiagnosisScoreDifferentialdiagnosisscoreisusedfortestingthesecondaryexpectedndings.Adifferentialdiagnosisisasystematicdiagnosticmethodusedtoidentifythepresenceofanentitywheremultiplealternativesarepossible[ 2 ].Intheinterimsurveyaftertheinterviewsession,eachparticipanthadtheopportunitytoidentifyvedifferentdifferentialdiagnosesbasedontheinformationtheparticipantwasabletoelicit.Thedifferentialdiagnosesareorderedbythemostlikelyrst.Dependingontherankofthecorrectdiagnosis,eachparticipantobtainsascorefrom0to5.Ifthecorrectdiagnosisistheparticipant'srstdifferentialdiagnosis,theparticipantobtainsascoreof5.Ifthecorrectdiagnosisistheparticipant'sfthdifferentialdiagnosis,theparticipantobtainsascoreof1.Ifthecorrectdiagnosisisnotanyoftheparticipant'sdifferentialdiagnosis,theparticipantobtainsascoreof0. 4.5.4NumberofQuestionsAskedbyParticipantsNumberofquestionsaskedbyparticipantstothemothersisusedfortestingthesecondaryexpectedndings.ItisimportanttocomparebetweenthenumberofquestionsaskedtotheEPandLEPmothers.TheLEPmothersneedtobeaskedmorequestionsthantheEPmotherstoelicitimportantinformationduetothelimitedabilityoftheLEPmotherstounderstand 34

PAGE 35

alloftheparticipant'squestions.Someoftheparticipant'squestionswouldneedtoberephrasedorbrokenintomoreelementaryquestions. 4.5.5SurveyResponsesForeachcase,participantshadtollinsurveysthathavequestionsaboutthatspeciccase.Herearethesurveyquestionsthatareofconcernfortheconductedstudy: Ihavethenecessaryskillsandknowledgetoelicitafocusedhistoryofthispatient(Q1). Iknewwhatpertinentinformationwasneededtomakethediagnosisforthispatient(Q2). ItwaseasytogettheinformationIwantedfromthemother(Q3). Themothergaveclearanswerstoallofmyquestions(Q4). Ididn'tneedtorephraseorclarifymyquestionstothemother(Q5). Ifeelcondentaboutthediagnosisbasedonthehistoryalone(Q6).Afternishingthetwocasesandthesurveys,participantshadtollinapostsurveyabouteachofthetwocasestheyinterviewed.Herearethesurveyquestionsthatareofconcernfortheconductedstudy: Elicitingthehistorywaseasy(Q7). Themotheralwaysunderstoodmyquestions(Q8). Themotherwasalwaysabletoanswermyquestions(Q9). Ineverhadtorephrasemyquestionsinordertogetinformationfromthemother(Q10).Participantshadtorespondtothosequestionsoncefortherstpatientandonceforthesecondpatient.Allofthepreviousquestionshavea5-pointLikertscaleasfollows: StronglyDisagree:1 Disagree:2 35

PAGE 36

NeitherAgreenorDisagree:3 Agree:4 StronglyAgree:5ParticipantsresponsestothesurveyquestionstestiftherewasadifferencebetweentheEPandtheLEPmothersregardingtheclarityoftheiranswersandtheirEnglishprociency. 36

PAGE 37

CHAPTER5STUDYRESULTSANDDISCUSSION 5.1PrimaryHypothesis1andPercentageofIdentiedDiscoveriesWeusedindependentfactorialANOVAwiththreeindependentfactors:Englishprociency(EP,LEP),caseorder(rst,second)andcasetype(Heart,Lung)foranalyzingthestudyresultsfortherstgroup.TherstgroupcontainstheparticipantsthatinterviewedbothEPandLEPmothers(23participants). 5.1.1ClassADiscoveries 5.1.1.1Three-wayinteractioneffectTherewasnosignicantthree-wayinteractioneffectbetweenEnglishprociency,casetypeandcaseorder(F1,38=0.066,p=0.798).Asaresult,thethree-wayinteractioneffectwasexcludedfromthemodel. 5.1.1.2Two-wayinteractioneffectTherewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcasetype(F1,39=0.005,p=0.946)orbetweencaseorderandcasetype(F1,39=0.01,p=0.921).Asaresultsthosetwo-wayinteractioneffectswereexcludedfromthemodel.Therewasasignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcaseorder(F1,41=10.116,p=0.003)whichneededafurtherinvestigation.FurtherinvestigationwascarriedthroughpairwisecomparisonsbetweenEnglishprociencyandcaseorderandthefollowingwasnoticed(Figure 5-1 ): WhentheEPmotherswereinterviewedaftertheLEPmothers,thepercentageofclassAdiscoveriesthatparticipantsidentiedfromtheEPmotherswassignicantlyhigherthanthepercentageofclassAdiscoveriesthatparticipantsidentiedfromtheEPmotherswheninterviewedbeforetheLEPmothers(F1,41=4.634,p=0.037).Thisindicatesthattherewasalearningeffect.WhentheEPmotherswereinterviewedaftertheLEPmothers,participantslearnedfromtheLEPmothersintherstinterviewandhencetheparticipantswereabletoelicitmoreinformationthantheydidwhentheEPmothersweretheirrstexperience. 37

PAGE 38

WhentheLEPmotherswereinterviewedbeforetheEPmothers,thepercentageofclassAdiscoveriesthatparticipantsidentiedfromtheLEPmotherswassignicantlyhigherthanthepercentageofclassAdiscoveriesthatparticipantsidentiedfromtheLEPmotherswheninterviewedaftertheEPmothers(F1,41=5.489,p=0.024).WhentheLEPmotherswereinterviewedbeforetheEPmothers,itwaspossiblethatparticipantshadlowexpectationsforthesystemsotheydidtheirbesttoelicitinformationfromtheLEPmothers.Ontheotherhand,whentheEPmotherswereinterviewedrst,participantshadhighexpectationsforthenextinterview.ThehighexpectationsforthenextinterviewcausedtheparticipantstoexpectfromtheLEPmotherswhattheyexperiencedfromtheEPmotherswhichwaswrongbecauseelicitinginformationfromtheLEPmotherswasharder.HenceparticipantselicitedsignicantlylessinformationfromtheLEPmothersthanwheninterviewedrst. Figure5-1. TheinteractioneffectbetweenEnglishprociencyandcaseorderforclassAdiscoveries. 5.1.1.3MaineffectsDuetothesignicantinteractioneffectbetweenEnglishprociencyandcaseorder,thedatawasdividedintotwogroupstogetridoftheordereffect.OnegrouphasthedataforparticipantsthatinterviewedanEPmotherfollowedbyanLEPmother.The 38

PAGE 39

othergrouphasthedataforparticipantsthatinterviewedanLEPmotherfollowedbyanEPmother.Englishprociency:FortheparticipantswhointerviewedanEPmotherfollowedbyanLEPmother,thepercentageofclassAdiscoveriesthatparticipantsidentiedfromtheEPmotherswassignicantlylargerthanthepercentageparticipantsidentiedfromtheLEPmothers(F1,19=9.110,p=0.007).FortheparticipantswhointerviewedanLEPmotherfollowedbyanEPmother,thepercentageofclassAdiscoveriesthatparticipantsidentiedfromtheEPmotherswassignicantlylargerthanthepercentageparticipantsidentiedfromtheLEPmothers(F1,21=11.447,p=0.003)(Figure 5-2 ). Figure5-2. ResultsforpercentageofidentiedclassAdiscoveriesfromtheEPandtheLEPmothers. Forthetwogroups,theresultsagreewithPrimaryHypothesis1.ItwasexpectedthatinterviewingtheEPmotherswouldbeeasierthaninterviewingtheLEPmothers.InterviewingtheLEPmotherswashardduetothelanguagebarriersthatpreventedthe 39

PAGE 40

LEPmothersfromunderstandingsomeoftheparticipants'questionsandhencenotbeingabletoanswerthosequestions.WhentheLEPmotherunderstandsaquestion,shemaygiveananswerthathaslessinformationthantheEPmotherwouldgive.ForexampletheEPmotherintheHeartCasewouldrespondtothequestionCanyoudescribehiscough?withHiscoughsoundswetanddeep,hedoesnotreallycoughanythingupthough.TheLEPmotherinthesameHeartCasewouldrespondtothesamequestionwithWell,formecoughiscough.TheresponseoftheLEPmotherwastheactualresponsewegotfromtheLEPfemalethatplayedtheroleofthemotherinthatcase.DuetothelackofenoughinformationinsomeoftheanswersoftheLEPmother,itmakessensethatparticipantswouldelicitlessinformationfromtheLEPmothersthantheywouldelicitfromtheEPmothersandhenceidentifyalowerpercentageofdiscoveries.Casetype:FortheparticipantswhointerviewedanEPmotherfollowedbyanLEPmother,thepercentageofclassAdiscoveriesthatparticipantsidentiedfromthemothersintheHeartCasewassignicantlylargerthanthepercentageparticipantsidentiedfromthemothersintheLungCase(F1,19=17.818,p<0.001).FortheparticipantswhointerviewedanLEPmotherfollowedbyanEPmother,thepercentageofclassAdiscoveriesthatparticipantsidentiedfromthemothersintheHeartCasewassignicantlylargerthanthepercentageparticipantsidentiedfromthemothersintheLungCase(F1,21=21.018,p<0.001)(Figure 5-3 ).Forthetwogroups,ThepercentageofclassAdiscoveriesthatparticipantsidentiedfromtheHeartCasewassignicantlylargerthanthepercentageparticipantsidentiedfromtheLungCasewhichindicatesthattheLungCasewasofahigherdifcultyregardingelicitingimportantinformation.However,therewasnointeractioneffectbetweentheEnglishprociencyandcasetypebecausewithineachcase,the 40

PAGE 41

Figure5-3. ResultsforpercentageofidentiedclassAdiscoveriesfromthemothersintheHeartandtheLangCases. numberofclassAdiscoveriesthattheparticipantsidentiedfromtheEPmotherswaslargerthanthenumberthattheparticipantsidentiedfromtheLEPmothers. 5.1.2ClassBDiscoveries 5.1.2.1Three-wayinteractioneffectTherewasnosignicantthree-wayinteractioneffectbetweenEnglishprociency,casetypeandcaseorder(F1,38=0.019,p=0.892).Asaresult,thethree-wayinteractioneffectwasexcludedfromthemodel. 5.1.2.2Two-wayinteractioneffectTherewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcasetype(F1,39=2.123,p=0.153)orbetweencaseorderandcasetype(F1,39=0.940,p=0.338).Asaresultsthosetwo-wayinteractioneffectswereexcludedfromthemodel. 41

PAGE 42

Therewasasignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcaseorder(F1,41=4.983,p=0.031)whichneededafurtherinvestigation.FurtherinvestigationwascarriedthroughpairwisecomparisonsbetweenEnglishprociencyandcaseorderandthefollowingwasnoticed(Figure 5-4 ): Figure5-4. TheinteractioneffectbetweenEnglishprociencyandcaseorderforclassBdiscoveries. TherewasnosignicantdifferencebetweenthepercentageofclassBdiscoveriesthatparticipantsidentiedfromtheEPmotherswhentheEPmotherswereinterviewedaftertheLEPmothersandthepercentageofclassBdiscoveriesthatparticipantsidentiedfromtheEPmotherswheninterviewedbeforetheLEPmothers(F1,41=0.822,p=0.37). WhentheLEPmotherswereinterviewedbeforetheEPmothers,thepercentageofclassBdiscoveriesthatparticipantsidentiedfromtheLEPmotherswassignicantlyhigherthanthepercentageofclassBdiscoveriesthatparticipantsidentiedfromtheLEPmotherswheninterviewedaftertheEPmothers(F1,41=5.058,p=0.03).ThissupportsourargumentaboutclassAdiscoveries. 42

PAGE 43

5.1.2.3MaineffectsDuetothesignicantinteractioneffectbetweenEnglishprociencyandcaseorder,thedatawasdividedintotwogroupstogetridoftheordereffect.OnegrouphasthedataforparticipantsthatinterviewedanEPmotherfollowedbyanLEPmother.TheothergrouphasthedataforparticipantsthatinterviewedanLEPmotherfollowedbyanEPmother.Englishprociency:FortheparticipantswhointerviewedanEPmotherfollowedbyanLEPmother,therewasnosignicantdifferencebetweenthepercentageofclassBdiscoveriesthatparticipantsidentiedfromtheEPmothersandthepercentageparticipantsidentiedfromtheLEPmothers(F1,19=0.544,p=0.470).FortheparticipantswhointerviewedanLEPmotherfollowedbyanEPmother,therewasnosignicantdifferencebetweenthepercentageofclassBdiscoveriesthatparticipantsidentiedfromtheEPmothersandthepercentageparticipantsidentiedfromtheLEPmothers(F1,21=0.384,p=0.542)(Figure 5-5 ).Casetype:FortheparticipantswhointerviewedanEPmotherfollowedbyanLEPmother,therewasnosignicantdifferencebetweenthepercentageofclassBdiscoveriesthatparticipantsidentiedfromthemothersintheHeartCaseandthepercentageparticipantsidentiedfromthemothersintheLungCase(F1,19=0.544,p=0.470).FortheparticipantswhointerviewedanLEPmotherfollowedbyanEPmother,therewasnosignicantdifferencebetweenthepercentageofclassBdiscoveriesthatparticipantsidentiedfromthemothersintheHeartCaseandthepercentageparticipantsidentiedfromthemothersintheLungCase(F1,21=0.384,p=0.542)(Figure 5-6 ).Ingeneral,amongallparticipants,theaveragepercentageofidentiedclassBdiscoverieswas28.8%.ThispercentageislowcomparedtotheaveragepercentageofidentiedclassAdiscoverieswhichwas40.6%. 43

PAGE 44

Figure5-5. ResultsforpercentageofidentiedclassBdiscoveriesfromtheEPandtheLEPmothers. ThelowaverageispossiblybecauseclassBdiscoveriesareofalowerimportancethanclassAdiscoveriesandhenceparticipantsdidnotfocusonthatkindofdiscoveries.Thelowaveragemakesithardertoseeasignicantdifference. 5.1.3ClassCDiscoveries 5.1.3.1Three-wayinteractioneffectTherewasnosignicantthree-wayinteractioneffectbetweenEnglishprociency,casetypeandcaseorder(F1,38=0.691,p=0.411).Asaresult,thethree-wayinteractioneffectwasexcludedfromthemodel. 5.1.3.2Two-wayinteractioneffectTherewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcasetype(F1,39=2.342,p=0.134)orbetweencaseorderandcasetype(F1,39=0.044,p=0.836).Therewasalsonosignicanttwo-wayinteractioneffect 44

PAGE 45

Figure5-6. ResultsforpercentageofidentiedclassBdiscoveriesfromthemothersintheHeartandtheLangCases. betweenEnglishprociencyandcaseorder(F1,39=1.398,p=0.244).Asaresultsalltwo-wayinteractioneffectswereexcludedfromthemodel. 5.1.3.3MaineffectsTherewasnosignicantdifferencebetweenthepercentageofclassCdiscoveriesfortheEPandtheLEPmothers(F1,42=0.879,p=0.354)orthemothersintheHeartCaseandthemotherintheLungCase(F1,42=0.879,p=0.354).TherewasalsonosignicantdifferencebetweenthepercentageofclassCdiscoveriesfortherstandthesecondinterviewedcases(F1,42=0.01,p=0.922)(Figure 5-7 ).Ingeneral,amongallparticipants,theaveragepercentageofidentiedclassCdiscoverieswas6.9%.ThisnumberisverylowcomparedtotheaveragepercentageofidentiedclassAdiscoverieswhichwas40.6%. 45

PAGE 46

Figure5-7. ResultsforpercentageofidentiedclassCdiscoveriesfromtheEPandtheLEPmothers,fromthemothersintheHeartandtheLangCasesandfromtherstandthesecondinterviewedmothers. ThelowaverageispossiblybecauseclassCdiscoveriesareofamuchlowerimportancethanclassAdiscoveriesandhenceparticipantsdidnotfocusonthatkindofdiscoveries.Thelowaveragemakesithardertoseeasignicantdifference. 5.2PrimaryHypothesis2andPercentageofIdentiedDiscoveriesWeusedindependentfactorialANOVAwithtwoindependentfactors:caseorder(rst,second)andcasetype(Heart,Lung)foranalyzingthestudyresultsforthesecondgroup.ThesecondgroupcontainstheparticipantsthatinterviewedtwoLEPmothers(8participants).TheresultsfornumberofclassA,BandCdiscoveriesregardingcaseorder(rstandsecond)aresummarizedinTable 5-1 46

PAGE 47

Table5-1. PercentageofidentieddiscoveriesfromtherstandthesecondinterviewedLEPmothersforeachclassofdiscoveries FirstLEPcasemean(st.dev)SecondLEPcasemean(st.dev)Pvalue PercentofidentiedclassAdiscoveries33.6%(6.7)36.6(10)0.248PercentofidentiedclassBdiscoveries33.76%(10.3)27.67(10.7)0.283PercentofidentiedclassCdiscoveries3.1%(8.8)7.8(9.3)0.319 5.2.1ClassADiscoveries 5.2.1.1Two-wayinteractioneffectTherewasnosignicanttwo-wayinteractioneffectbetweencasetypeandcaseorder(F1,12=1.525,p=0.241).Asaresult,thetwo-wayinteractioneffectwasexcludedfromthemodel. 5.2.1.2MaineffectsCaseorder:TherewasnosignicantdifferencebetweenthepercentageofclassAdiscoveriesidentiedfromtherstinterviewedLEPmothersandthesecondinterviewedLEPmothers(F1,13=1.466,p=0.248).ThisresultrejectsPrimaryHypothesis2aboutparticipantsimprovingtheircommunicationskillsfordealingwiththeLEPmothersafterusingthesystem.Onefactorthatmayhaveaffectedtheresultisthattherewereonly8participants(outof31)thatinterviewedtwoLEPmothersoneafteranother.Also,participantsmighthavelearnedfromthesysteminbothinterviewsandhencenosignicantdifferenceexistedbetweenthepercentageofidentieddiscoveriesinthetwointerviews.Casetype:ThepercentageofclassAdiscoveriesthatparticipantsidentiedfromthemothersintheHeartCasewassignicantlylargerthanthepercentageparticipantsidentiedfromthemothersintheLungCase(F1,13=27.784,p<0.001)(Figure 5-8 ).ThisagreeswithourargumentabouttheLungCasebeingmoredifcultthantheHeartCase. 47

PAGE 48

Figure5-8. ResultsforpercentageofidentiedclassAdiscoveriesfromthemothersintheHeartandtheLangCases(forparticipantsinthesecondgroup). 5.2.2ClassBDiscoveriesTherewasnosignicantdifferencebetweenthepercentageofclassBdiscoveriesthatparticipantsidentiedfromtheLEPmotherintherstcaseandtheLEPmotherinthesecondcase(F1,13=1.253,p=0.283). 5.2.3ClassCDiscoveriesTherewasnosignicantdifferencebetweenthepercentageofclassCdiscoveriesthatparticipantsidentiedfromtheLEPmotherintherstcaseandtheLEPmotherinthesecondcase(F1,13=1.073,p=0.319). 5.3PrimaryHypothesis3andTechniqueAdaptationtoRephrasingQuestionsTwelveparticipantsoutofthethirtyone(38.71%)mentionedthattheyusedrephrasingtoelicitinformationfromtheLEPmothers(TechniqueAdaptationScore=1).Table 5-2 showstheresponseoftwooftheseparticipantsaboutthetechniquestheyanticipatedtousetoelicitinformationfromanLEPpatientbeforeandaftertheir 48

PAGE 49

Table5-2. Techniqueadaptationtorephrasingquestionsfortwoparticipants. BeforetheinterviewwithanLEPmotherAftertheinterviewwithanLEPmother Participant1Useotherfamilymemberswhocaninterpret.Callaninterpreterservice.Usemyhandstogestureandtryanddescribemyquestions.Itriedrephrasingquestionswithsomesuccess.Ialsousedmorecommonwords,suchaspoopinsteadofstoolwhenthemotherdidnotunderstand.Participant2Itrytousesimplewords.ButiftheirlevelsofunderstandingofEnglishistoolow,Iwouldgetaninterpreteroruselanguageline.Ihaven'tbeeninthatsituationyet.Usingmoresimplewords.Forexample,shedidn'tunderstandthewordfussy.SoIworkedaroundthat. actualinterviewwiththeLEPmothers.Therestoftheparticipantsmighthaveusedrephrasingwithoutrealizingandhencethoseparticipantsdidnotmentionthat.Ingeneral,astrongermeasureisneededtobeusedtosupportPrimaryHypothesis3. 5.4SecondaryFindingsWeusedindependentfactorialANOVAwiththreeindependentfactors:Englishprociency(EP,LEP),caseorder(rst,second)andcasetype(Heart,Lung)foranalyzingthestudyresultsfortherstgroup.TherstgroupcontainstheparticipantsthatinterviewedbothEPandLEPmothers(23participants). 5.4.1DifferentialDiagnosisScore 5.4.1.1Three-wayinteractioneffectTherewasnosignicantthree-wayinteractioneffectbetweenEnglishprociency,casetypeandcaseorder(F1,38=1.865,p=0.18).Asaresult,thethree-wayinteractioneffectwasexcludedfromthemodel. 5.4.1.2Two-wayinteractioneffectTherewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcasetype(F1,39=0.534,p=0.469)orbetweenEnglishprociencyandcaseorder(F1,39=0.083,p=0.775).Asaresultsthosetwo-wayinteractioneffectswereexcludedfromthemodel. 49

PAGE 50

Therewasasignicanttwo-wayinteractioneffectbetweencasetypeandcaseorder(F1,41=5.193,p=0.028).Furtherinvestigationwascarriedthroughpairwisecomparisonsbetweencasetypeandcaseorderandthefollowingwasnoticed(Figure 5-9 ): Figure5-9. Theinteractioneffectbetweencasetypeandcaseorderforthedifferentialdiagnosisscore. FortheHeartCase,thedifferentialdiagnosisscorethattheparticipantsobtainedfromtheirsecondinterviewwassignicantlylargerthanthescorethattheparticipantsobtainedfromtheirrstinterview(F1,41=7.275,p=0.01).ThisresultpossiblyindicatesthattheparticipantslearnedfromtheLungCaseintherstinterviewandhenceobtainedahigherscorefortheirdifferentialdiagnosisregardingtheHeartCaseinthesecondinterview. FortheLungCase,therewasnosignicantdifferencebetweenthedifferentialdiagnosisscorethatparticipantsobtainedfromtheirrstandsecondinterviews(F1,41=0.274,p=0.603).ItispossiblethattheHeartCaseintherstinterviewbiasedthedifferentialdiagnosisfortheLungcase(whichprovedtobeharderthantheHeartCase)inthesecondinterview. 50

PAGE 51

5.4.1.3MaineffectsDuetothesignicantinteractioneffectbetweencasetypeandcaseorder,thedatawasdividedintotwogroupstogetridoftheordereffect.OnegrouphasthedataforparticipantsthatinterviewedanEPmotherfollowedbyanLEPmother.TheothergrouphasthedataforparticipantsthatinterviewedanLEPmotherfollowedbyanEPmother.Englishprociency:FortheparticipantswhointerviewedanEPmotherfollowedbyanLEPmother,therewasnosignicantdifferencebetweenthedifferentialdiagnosisscorethattheparticipantsobtainedafterinterviewingtheEPmothersandthescorethattheparticipantsobtainedafterinterviewingtheLEPmothers(F1,19=2.792,p=0.111).FortheparticipantswhointerviewedanLEPmotherfollowedbyanEPmother,therewasnosignicantdifferencebetweenthedifferentialdiagnosisscorethattheparticipantsobtainedafterinterviewingtheEPmothersandthescorethattheparticipantsobtainedafterinterviewingtheLEPmothers(F1,21=0.214,p=0.649)(Figure 5-10 ).However,thisresultdoesnotmeanthatlanguagebarriershavenoeffectonphysician'sdifferentialdiagnosisbecausetherewasafactorthataffectedtheresults.ThefactorthataffectedtheresultswasthedifcultyoftheHeartandLungCases.ThechosenHeartandLungCasesarerarelypresentedinclinicaltraining.Presentingthesecasesthroughthevirtualhumansisstillgoodfortrainingpurposes.Ingeneral,thedifferentialdiagnosisscoresthattheparticipantsgotforboththeEPandtheLEPmotherswereverylowasmostoftheparticipantsscoredzero(Figure 5-11 ).Thissupportsourargumentthattheresultdoesnotmeanthatlanguagebarriershavenoeffectondifferentialdiagnosis.Theresultsmerelyindicatesthatthecaseswepresentedtotheparticipantswerenotcapableofshowingasignicantdifference.Casetype:FortheparticipantswhointerviewedanEPmotherfollowedbyanLEPmother,thedifferentialdiagnosisscorethattheparticipantsobtainedafter 51

PAGE 52

Figure5-10. ResultsofdifferentialdiagnosisscoreforEnglishprociency(EP,LEP). interviewingthemothersintheHeartCasewassignicantlylargerthanthescorethattheparticipantsobtainedafterinterviewingthemothersintheLungCase(F1,19=5.142,p=0.035).FortheparticipantswhointerviewedanLEPmotherfollowedbyanEPmother,therewasnosignicantdifferencebetweenthedifferentialdiagnosisscorethattheparticipantsobtainedafterinterviewingthemothersintheHeartCaseandthescorethattheparticipantsobtainedafterinterviewingthemothersintheLungCase(F1,21=0.214,p=0.649)(Figure 5-12 ).ThispossiblyindicatesthatidentifyingthediagnosisfortheLungCasewasharderthanidentifyingthediagnosisfortheHeartCasebutthereisnoobviousreason,whywasthatsignicantfortherstgrouponly.Thismightbeduetosomedifferencesbetweentheparticipantsinthetwogroups. 52

PAGE 53

Figure5-11. Histogramfordifferentialdiagnosisscores 5.4.2NumberofQuestions 5.4.2.1Three-wayinteractioneffectTherewasnosignicantthree-wayinteractioneffectbetweenEnglishprociency,casetypeandcaseorder(F1,38=0.586,p=0.449).Asaresult,thethree-wayinteractioneffectwasexcludedfromthemodel. 5.4.2.2Two-wayinteractioneffectTherewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcasetype(F1,39=3.746,p=0.06)orbetweencaseorderandcasetype(F1,39=0.122,p=0.729).Asaresultsthosetwo-wayinteractioneffectswereexcludedfromthemodel.Therewasasignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcaseorder(F1,41=4.227,p=0.046). 53

PAGE 54

Figure5-12. Resultsofdifferentialdiagnosisscoreforcasetype(Heart,Lung). 5.4.2.3MaineffectsDuetothesignicantinteractioneffectbetweenEnglishprociencyandcaseorder,thedatawasdividedintotwogroupstogetridoftheordereffect.OnegrouphasthedataforparticipantsthatinterviewedanEPmotherfollowedbyanLEPmother.TheothergrouphasthedataforparticipantsthatinterviewedanLEPmotherfollowedbyanEPmother.Englishprociency:FortheparticipantswhointerviewedanEPmotherfollowedbyanLEPmother,therewasnosignicantdifferencebetweenthenumberofquestionsthatparticipantsaskedtotheEPmothersandthenumberofquestionsthatparticipantsaskedtotheLEPmothers(F1,19=0.333,p=0.571).FortheparticipantswhointerviewedanLEPmotherfollowedbyanEPmother,therewasnosignicantdifferencebetweenthenumberofquestionsthatparticipantsaskedtothe 54

PAGE 55

Figure5-13. ResultsfornumbersofquestionsaskedtotheEPandtheLEPmothers. EPmothersandthenumberofquestionsthatparticipantsaskedtotheLEPmothers(F1,21=0.574,p=0.457)(Figure 5-13 ).Casetype:FortheparticipantswhointerviewedanEPmotherfollowedbyanLEPmother,therewasnosignicantdifferencebetweenthenumberofquestionsthatparticipantsaskedtothemothersintheHeartCaseandthenumberofquestionsthatparticipantsaskedtothemothersintheLungCase(F1,19=0.539,p=0.472).FortheparticipantswhointerviewedanLEPmotherfollowedbyanEPmother,therewasnosignicantdifferencebetweenthenumberofquestionsthatparticipantsaskedtothemothersintheHeartCaseandthenumberofquestionsthatparticipantsaskedtothemothersintheLungCase(F1,21=0.162,p=0.691)(Figure 5-14 ).Theinsignicantresultsarepossiblyduetothefactthatallquestionsaskedbytheparticipantweretakenintoaccountincludingrepeatedquestions.Repeatedquestions 55

PAGE 56

Figure5-14. ResultsfornumbersofquestionsaskedtothemothersintheHeartandtheLungCases. weresometimesduetothelimitationofthesystemtorecognizethosequestionsbutsomeparticipantsinsistedonaskingthesamequestionmorethanonceindifferentways.Itwouldbebettertobreakdowntheaskedquestionsintocategoriestocapturesignicantvariations. 5.4.3IHavetheNecessarySkillsandKnowledgetoElicitaFocusedHistoryofThisPatient(Q1) 5.4.3.1Three-wayinteractioneffectTherewasnosignicantthree-wayinteractioneffectbetweenEnglishprociency,casetypeandcaseorder(F1,38=0.186,p=0.669).Asaresult,thethree-wayinteractioneffectwasexcludedfromthemodel. 56

PAGE 57

5.4.3.2Two-wayinteractioneffectTherewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcasetype(F1,39=0.248,p=0.621)orbetweencaseorderandcasetype(F1,39=1.255,p=0.269).Asaresultsthosetwo-wayinteractioneffectswereexcludedfromthemodel.Therewasasignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcaseorder(F1,41=7.354,p=0.01). 5.4.3.3MaineffectsDuetothesignicantinteractioneffectbetweenEnglishprociencyandcaseorder,thedatawasdividedintotwogroupstogetridoftheordereffect.OnegrouphasthedataforparticipantsthatinterviewedanEPmotherfollowedbyanLEPmother.TheothergrouphasthedataforparticipantsthatinterviewedanLEPmotherfollowedbyanEPmother.Englishprociency:FortheparticipantswhointerviewedanEPmotherfollowedbyanLEPmother,therewasnosignicantdifferencebetweenthelevelofinterviewingcondenceofparticipantswhointerviewedtheEPmothersandforthoseparticipantswhointerviewedtheLEPmothers(F1,19=0.122,p=0.731).FortheparticipantswhointerviewedanLEPmotherfollowedbyanEPmother,therewasnosignicantdifferencebetweenthelevelofinterviewingcondenceofparticipantswhointerviewedtheEPmothersandforthoseparticipantswhointerviewedtheLEPmothers(F1,21=0.574,p=0.457)(Figure 5-15 ).ItisobviousfromthesurveyresultsthattheparticipantsingeneralwerecondantaboutabouttheirskillsandknowledgeforelicitingafocusedhistoryfortheirpatientsregardlessofthelevelofEnglishprociencyoftheinterviewedmother. 57

PAGE 58

Figure5-15. ResultsforQ1(Ihavethenecessaryskillsandknowledgetoelicitafocusedhistoryofthispatient). 5.4.4IKnewWhatPertinentInformationWasNeededtoMaketheDiagnosisforThisPatient(Q2) 5.4.4.1Three-wayinteractioneffectTherewasnosignicantthree-wayinteractioneffectbetweenEnglishprociency,casetypeandcaseorder(F1,38=0.155,p=0.696).Asaresult,thethree-wayinteractioneffectwasexcludedfromthemodel. 5.4.4.2Two-wayinteractioneffectTherewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcasetype(F1,39=0.033,p=0.857)orbetweencaseorderandcasetype(F1,39=2.649,p=0.112).Therewasalsonosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcaseorder(F1,39=2.649,p=0.112).Asaresultsalltwo-wayinteractioneffectswereexcludedfromthemodel. 58

PAGE 59

5.4.4.3MaineffectsEnglishprociency:TherewasnosignicantdifferencebetweentheknowledgeofinformationneededfordiagnosisforparticipantsthatinterviewedtheEPmothersandforthoseparticipantswhointerviewedtheLEPmothers(F1,38=0.112,p=0.740)(Figure 5-16 ).ItisobviousfromthesurveyresultsthattheparticipantsingeneralwerecondantabouttheinformationneededtomakethediagnosisfortheirpatientsregardlessofthelevelofEnglishprociencyoftheinterviewedmother. Figure5-16. ResultsforQ2(Iknewwhatpertinentinformationwasneededtomakethediagnosisforthispatient). 5.4.5ItWasEasytoGettheInformationIWantedFromtheMother(Q3) 5.4.5.1Three-wayinteractioneffectTherewasnosignicantthree-wayinteractioneffectbetweenEnglishprociency,casetypeandcaseorder(F1,38=0.494,p=0.486).Asaresult,thethree-wayinteractioneffectwasexcludedfromthemodel. 59

PAGE 60

5.4.5.2Two-wayinteractioneffectTherewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcaseorder(F1,39=0.031,p=0.86)orbetweencaseorderandcasetype(F1,39=0.014,p=0.907).Asaresultsthosetwo-wayinteractioneffectswereexcludedfromthemodel.Therewasasignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcasetype(F1,41=6.769,p=0.013). 5.4.5.3MaineffectsDuetothesignicantinteractioneffectbetweenEnglishprociencyandcasetype,thedatawasdividedintotwogroupstogetridofthecasetypeeffect.OnegrouphasthedataforparticipantsthatinterviewedanEPmotherfortheHeartCaseandanLEPmotherfortheLungCase.TheothergrouphasthedataforparticipantsthatinterviewedanEPmotherfortheLungCaseandanLEPmotherfortheHeartCase.Englishprociency:participantsthoughtthattherewasnosignicantdifferencebetweenelicitinginformationfromtheEPmothersintheHeartCaseandelicitinginformationfromtheLEPmothersintheLungCase(F1,21=1.679,p=0.209).participantsthoughtthatelicitinginformationfromtheEPmothersintheLungCasewassignicantlyeasierthanelicitinginformationfromtheLEPmothersintheHeartCase(F1,21=7.174,p=0.015)(Figure 5-17 ).TheseresultsmighthavebeenbeduetoanunintentionaldifferencebetweenthetwoEPmotherswhichmighthavecausedelicitinginformationfromtheEPmotherintheHeartCasetobealmostasdifcultaselicitinginformationfromtheLEPmotherintheLungCase.Oritmightbeduetosomerandomchoicesfortheparticipants. 60

PAGE 61

Figure5-17. ResultsforQ3(ItwaseasytogettheinformationIwantedfromthemother). 5.4.6TheMotherGaveClearAnswerstoAllofMyQuestions(Q4) 5.4.6.1Three-wayinteractioneffectTherewasnosignicantthree-wayinteractioneffectbetweenEnglishprociency,casetypeandcaseorder(F1,38=0.001,p=0.970).Asaresult,thethree-wayinteractioneffectwasexcludedfromthemodel. 5.4.6.2Two-wayinteractioneffectTherewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcaseorder(F1,39=0.525,p=0.473)orbetweencaseorderandcasetype(F1,39=0.641,p=0.428).Asaresultsthosetwo-wayinteractioneffectswereexcludedfromthemodel.Therewasasignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcasetype(F1,41=5.194,p=0.028). 61

PAGE 62

5.4.6.3MaineffectsDuetothesignicantinteractioneffectbetweenEnglishprociencyandcasetype,thedatawasdividedintotwogroupstogetridofthecasetypeeffect.OnegrouphasthedataforparticipantsthatinterviewedanEPmotherfortheHeartCaseandanLEPmotherfortheLungCase.TheothergrouphasthedataforparticipantsthatinterviewedanEPmotherfortheLungCaseandanLEPmotherfortheHeartCase.Englishprociency:participantsthoughtthattheanswersoftheEPmothersintheHeartCaseweresignicantlyclearerthantheanswersoftheLEPmothersintheLungCase(F1,21=15.464,p=0.001).ParticipantsthoughtthattherewasnosignicantdifferencebetweentheclarityoftheanswersoftheEPmothersintheLungCaseandtheclarityoftheanswersoftheLEPmothersintheHeartCase(F1,19=0.958,p=0.34)(Figure 5-18 ). Figure5-18. ResultsforQ4(Themothergaveclearanswerstoallofmyquestions). 62

PAGE 63

TheseresultsmighthavebeenbeduetoanunintentionaldifferencebetweenthetwoLEPmotherswhichmighthavecausedtheanswersoftheLEPmotherintheHeartCasetobealmostasclearastheanswersoftheEPmotherintheLungCase.Oritmightbeduetosomerandomchoicesfortheparticipants. 5.4.7IDidNotNeedtoRephraseorClarifyMyQuestionstotheMother(Q5) 5.4.7.1Three-wayinteractioneffectTherewasnosignicantthree-wayinteractioneffectbetweenEnglishprociency,casetypeandcaseorder(F1,38=0.24,p=0.627).Asaresult,thethree-wayinteractioneffectwasexcludedfromthemodel. 5.4.7.2Two-wayinteractioneffectTherewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcaseorder(F1,39=1.277,p=0.265)orbetweencaseorderandcasetype(F1,39=0.005,p=0.944).Asaresultsthosetwo-wayinteractioneffectswereexcludedfromthemodel.Therewasasignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcasetype(F1,41=4.520,p=0.04). 5.4.7.3MaineffectsDuetothesignicantinteractioneffectbetweenEnglishprociencyandcasetype,thedatawasdividedintotwogroupstogetridofthecasetypeeffect.OnegrouphasthedataforparticipantsthatinterviewedanEPmotherfortheHeartCaseandanLEPmotherfortheLungCase.TheothergrouphasthedataforparticipantsthatinterviewedanEPmotherfortheLungCaseandanLEPmotherfortheHeartCase.Englishprociency:ParticipantsthoughtthattherewasnosignicantdifferencebetweentheneedofrephrasingfortheEPmothersintheHeartCaseandtheneedofrephrasingfortheLEPmothersintheLungCase(F1,21=0.645,p=0.431).Participantsthoughtthattherewasnosignicantdifferencebetweentheneedof 63

PAGE 64

rephrasingfortheEPmothersintheLungCaseandtheneedofrephrasingfortheLEPmothersintheHeartCase(F1,19=0.262,p=0.614)(Figure 5-19 ). Figure5-19. ResultsforQ5(Ididnotneedtorephraseorclarifymyquestionstothemother). TheinsignicantdifferencemighthavebeenbeduetotheneedforrephrasingingeneralforbothEPandLEPmothersbecauseofthelimitationsinthesystemdatabase. 5.4.8IFeelCondentAbouttheDiagnosisBasedontheHistoryAlone(Q6) 5.4.8.1Three-wayinteractioneffectTherewasnosignicantthree-wayinteractioneffectbetweenEnglishprociency,casetypeandcaseorder(F1,38=0.907,p=0.347).Asaresult,thethree-wayinteractioneffectwasexcludedfromthemodel. 5.4.8.2Two-wayinteractioneffectTherewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcaseorder(F1,39=0.417,p=0.522)orbetweencaseorderandcasetype 64

PAGE 65

(F1,39=2.271,p=0.14).Asaresultsthosetwo-wayinteractioneffectswereexcludedfromthemodel.Therewasasignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcasetype(F1,41=7.935,p=0.007). 5.4.8.3MaineffectsDuetothesignicantinteractioneffectbetweenEnglishprociencyandcasetype,thedatawasdividedintotwogroupstogetridofthecasetypeeffect.OnegrouphasthedataforparticipantsthatinterviewedanEPmotherfortheHeartCaseandanLEPmotherfortheLungCase.TheothergrouphasthedataforparticipantsthatinterviewedanEPmotherfortheLungCaseandanLEPmotherfortheHeartCase.Englishprociency:TherewasnosignicantdifferencebetweenthecondenceofdiagnosisforparticipantsthatinterviewedtheEPmothersintheHeartCaseandthecondenceofdiagnosisforparticipantsthatinterviewedtheLEPmothersintheLungCase(F1,21=0.074,p=0.789).TherewasnosignicantdifferencebetweenthecondenceofdiagnosisforparticipantsthatinterviewedtheEPmothersintheLungCaseandthecondenceofdiagnosisforparticipantsthatinterviewedtheLEPmothersintheHeartCase(F1,19=0.252,p=0.622)(Figure 5-20 ).TheseresultsshowsthatingeneralthecondenceaboutthediagnosiswaslowregardlessthelevelofEnglishprociencyofthemothers.Theseresultsareconsistentwiththelowdiagnosisscorethattheparticipantsobtainedingeneral. 5.4.9ElicitingtheHistoryWasEasy(Q7) 5.4.9.1Three-wayinteractioneffectTherewasnosignicantthree-wayinteractioneffectbetweenEnglishprociency,casetypeandcaseorder(F1,38=0.403,p=0.529).Asaresult,thethree-wayinteractioneffectwasexcludedfromthemodel. 65

PAGE 66

Figure5-20. ResultsforQ6(Ifeelcondentaboutthediagnosisbasedonthehistoryalone). 5.4.9.2Two-wayinteractioneffectTherewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcasetype(F1,39=1.805,p=0.187)orbetweencaseorderandcasetype(F1,39=0.495,p=0.486).Asaresultsthosetwo-wayinteractioneffectswereexcludedfromthemodel.Therewasasignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcaseorder(F1,41=8.114,p=0.007). 5.4.9.3MaineffectsDuetothesignicantinteractioneffectbetweenEnglishprociencyandcaseorder,thedatawasdividedintotwogroupstogetridoftheordereffect.OnegrouphasthedataforparticipantsthatinterviewedanEPmotherfollowedbyanLEPmother.The 66

PAGE 67

othergrouphasthedataforparticipantsthatinterviewedanLEPmotherfollowedbyanEPmother.Englishprociency:FortheparticipantswhointerviewedanEPmotherfollowedbyanLEPmother,therewasnosignicantdifferencebetweenelicitinginformationfromtheEPmothersandelicitinginformationfromtheLEPmothers(F1,19=0.922,p=0.349).FortheparticipantswhointerviewedanLEPmotherfollowedbyanEPmother,elicitinginformationfromtheEPmotherswasnosignicantlyeasierthanelicitinginformationfromtheLEPmothers(F1,21=4.565,p=0.045)(Figure 5-21 ).Thereisnologicalexplanation,whythesignicanceexistsonlyinthesecondgroup. Figure5-21. ResultsforQ7(Elicitingthehistorywaseasy). 67

PAGE 68

5.4.10TheMotherAlwaysUnderstoodMyQuestions(Q8) 5.4.10.1Three-wayinteractioneffectTherewasnosignicantthree-wayinteractioneffectbetweenEnglishprociency,casetypeandcaseorder(F1,38=0.406,p=0.528).Asaresult,thethree-wayinteractioneffectwasexcludedfromthemodel. 5.4.10.2Two-wayinteractioneffectTherewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcaseorder(F1,39=0.0,p=1.0)orbetweencaseorderandcasetype(F1,39=0.085,p=0.772).Also,therewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcasetype(F1,39=0.085,p=0.772).Asaresultsalltwo-wayinteractioneffectswereexcludedfromthemodel. 5.4.10.3MaineffectsEnglishprociency:Participantsthoughtthattheabilityofunderstandingtheparticipants'questionsfortheEPmotherswassignicantlyhigherthantheabilityofunderstandingtheparticipants'questionsfortheLEPmothers(F1,42=4.366,p=0.043)(Figure 5-22 ).ThisresultsupportsthecapabilityofthevirtualhumanstosimulatehumanswithdifferentlevelsofEnglishprociencyasingeneralEPhumansareabletounderstandquestionsaskedinEnglishsignicantlybetterthanLEPhumans. 5.4.11TheMotherWasAlwaysAbletoAnswerMyQuestions(Q9) 5.4.11.1Three-wayinteractioneffectTherewasnosignicantthree-wayinteractioneffectbetweenEnglishprociency,casetypeandcaseorder(F1,38=0.216,p=0.645).Asaresult,thethree-wayinteractioneffectwasexcludedfromthemodel. 5.4.11.2Two-wayinteractioneffectTherewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcaseorder(F1,39=0.932,p=0.34)orbetweencaseorderandcasetype 68

PAGE 69

Figure5-22. ResultsforQ8(Themotheralwaysunderstoodmyquestions). (F1,39=0.476,p=0.494).Also,therewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcasetype(F1,39=1.541,p=0.222).Asaresultsalltwo-wayinteractioneffectswereexcludedfromthemodel. 5.4.11.3MaineffectsEnglishprociency:Participantsthoughtthattheabilityofansweringtheparticipants'questionswassignicantlyhigherfortheEPmothersthanfortheLEPmothers(F1,42=5.706,p=0.021)(Figure 5-23 ).ThisresultsupportsthecapabilityofthevirtualhumanstosimulatehumanswithdifferentlevelsofEnglishprociencyasingeneralEPhumansareabletoanswerquestionsthatareaskedinEnglishsignicantlybetterthanLEPhumans. 69

PAGE 70

Figure5-23. ResultsforQ9(Themotherwasalwaysabletoanswermyquestions). 5.4.12INeverHadtoRephraseMyQuestionsinOrdertoGetInformationFromtheMother(Q10) 5.4.12.1Three-wayinteractioneffectTherewasnosignicantthree-wayinteractioneffectbetweenEnglishprociency,casetypeandcaseorder(F1,38=0.205,p=0.654).Asaresult,thethree-wayinteractioneffectwasexcludedfromthemodel. 5.4.12.2Two-wayinteractioneffectTherewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcaseorder(F1,39=0.292,p=0.592)orbetweencaseorderandcasetype(F1,39=1.880,p=0.178).Also,therewasnosignicanttwo-wayinteractioneffectbetweenEnglishprociencyandcasetype(F1,39=0.913,p=0.345).Asaresultsalltwo-wayinteractioneffectswereexcludedfromthemodel. 70

PAGE 71

5.4.12.3MaineffectsEnglishprociency:ParticipantsthoughtthattherewasnosignicantdifferencebetweentheneedofrephrasingquestionsfortheEPmothersandtheneedofrephrasingquestionsfortheLEPmothers(F1,42=0.662,p=0.421)(Figure 5-24 ). Figure5-24. ResultsforQ10(Ineverhadtorephrasemyquestionsinordertogetinformationfromthemother). TheinsignicantdifferencemighthavebeenbeduetotheneedforrephrasingingeneralforbothEPandLEPmothersbecauseofthelimitationsinthesystemdatabase. 71

PAGE 72

CHAPTER6LIMITATIONANDFUTUREWORK 6.1Limitations 6.1.1LimitedDatabaseThesystemhas220responsesto1200differentquestionsintheEPscriptsand260responsesto1200differentquestionsintheLEPscripts.Thislimitednumberofquestionsinthesystemdatabasemayhavehadanegativeeffectontheperformanceoftheparticipants.Thislimitednumberofquestionsmayhavecausedtheparticipantstorephrasetheirquestionstothemothers.Consequently,theparticipantswerenotabletodistinguishbetweenrephrasingforthesystemtorecognizethequestionandrephrasingfortheLEPmotherstounderstandthequestion. 6.1.2DifcultyofMedicalCasesTwomedicalcaseswereusedinthestudy:HeartCaseandLungCase.Bothmedicalcaseswereofhighdifcultyforparticipantstocometothecorrectdiagnosis.Itwouldbebettertohavemorecommonmedicalcases. 6.1.3LimitedNumberofParticipantsThenumberofparticipantswasreasonablyenoughtodrawconclusionsabouttheeffectsofEnglishprociencyonmedicalinterviewingoutcomes.However,someofthestatisticalconclusionsmightbestrengthenedbyalargersampleofparticipants. 6.2FutureWork 6.2.1ImprovingtheDatabaseThenumberofquestionsinthescriptdatabasecanbeincreasedbyaddingalltheunmatchedquestionsoftheparticipantstothedatabase.Byaddingallthosequestionstothedatabase,themotherswillbeabletorespondtoalargernumberofquestionswhichwillimprovetheexperienceoftheuserwiththesystem. 72

PAGE 73

6.2.2HavingtheSystemOnlineTransferringthesystemintoanewplatformsoastohaveitavailableonlinecanincreasethenumberofparticipantsandhenceimprovetheresults. 6.2.3GivingImmediateFeedbackThesystemimprovesthecapabilitiesofhealthcareprovidersinelicitinginformationfromanLEPmotherbygivingtheprovidersresponsesfromthevirtualLEPmotherthatreectherunderstandingoftheirquestions.Forexample,ananswerlikeIdon'tknowwhatisdischargetellsprovidersthatthemothercouldnotunderstandtheirquestion.Inordertoelicittheinformationproviderswant,providersneedtorephrasetheirquestionsandtousesimplerwords.Currentlythesystemgivestheproviderstheimpressionthattheyneedtorephrasetheirquestionsbutitdoesnottellprovidersexplicitlywhatiswrongwiththeirquestions.Forfuturework,weareplanningtomakethesystemmoreeducationalwhendealingwithLEPcasesbygivinghealthcareprovidersimmediatefeedbackaboutwhatwaswrongwiththeirquestionsandwhattheyneedtoaskinstead.Forthepreviousexample,itwouldbenicethatthesystemcouldtellprovidersthatdischargeishardforanLEPparenttounderstand.Moreover,itwouldalsotellhealthcareprovidersthattheycanreplacetheworddischargebyuidsforinstance. 6.2.4PossibleStudies 6.2.4.1OrdereffectonLEPpatientsWhentheEPmotherswereinterviewedaftertheLEPmothers,thepercentageofclassAdiscoveriesthatparticipantsidentiedfromtheEPmotherswassignicantlyhigherthanthepercentageofclassAdiscoveriesthatparticipantsidentiedfromtheEPmotherswheninterviewedbeforetheLEPmothers.Thisresultmakessensebecauseofthelearningeffect.However,whentheLEPmotherswereinterviewedbeforetheEPmothers,thepercentageofclassAdiscoveriesthatparticipantsidentiedfromtheLEPmotherswassignicantlyhigherthanthepercentageofclassAdiscoveriesthat 73

PAGE 74

participantsidentiedfromtheLEPmotherswheninterviewedaftertheEPmothers.Furtherstudiesareneedstobeconductedtounderstandthepreviousresult.WouldparticipantsalwaysperformworsewithanLEPparentwheninterviewedsecondlyorwasitbecausetheLEPmotherswereinterviewedafterEPmothers?Theresultsfromthatfuturestudycanbeveryusefulformedicaltraining.MedicaltrainingmayneedtoalwaysstartonLEPpatientswhichwillalsobenetEPpatients.IfmedicaltrainingisinitializedonEPpatients,thatmighthindertheperformancewithLEPpatientslater. 6.2.4.2ImprovingthequalityofhealthcareforLEPpatientsFurtherstudiesareneededtobeconductedtoinvestigatethepossibilitythatoursystemhastheabilitytoimprovethequalityofhealthcarethatLEPpatientsgetbypossiblyfollowingupwithourthirtyoneparticipantsandassessingtheircurrentperformancewithLEPpatients.Alsonewstudiescanbeconductedtocheckforimprovementinparticipants'performancewhentheyuseoursystemovertimeespeciallyafterhavingthesystemdeployedonline. 74

PAGE 75

CHAPTER7CONCLUSIONInthisthesis,weintroducedanovelvirtualenvironmentforassessingthequalityofhealthcarethatLEPpatientsgetincomparisonwiththequalityofhealthcarethatEPpatientsget.Wedemonstratedthroughauserstudythattheeducationalbenetsthatmedicalstudentsandresidentsgotfrominteractingwithvirtualmotherswereasgoodasthosegainedfromdealingwithreal-lifecases.Aswell,themedicalstafflearnedhowtorephrasetheirquestionstoovercomemisunderstandingandconfusionastheydealtwithLEPmothers.TheresultsdrawnfromtheproposedsystemarehighlyencouragingandshowagreatpotentialforusingvirtualhumansforsimulatingLEPhumanstoassessthequalityofhealthcareinLEPgroups.Theusefulnessoftheproposedsystemcanbefurtherimprovedbyenlargingtheassociatedmedicalscriptdatabaseanddeployingthesystemontheweb.UsingvirtualhumansforsimulatingLEPhumansmaynotonlybeusedinthemedicaleld.UsingvirtualhumansforsimulatingLEPhumansmaybeusedinthemilitaryoranyothereld.FurtherstudiesareneededwithdifferentmedicalcasesthataremorecommonthantheHeartandLungcasesthatweusedinourstudy.AlsofurtherstudiesareneededtocheckforordereffectasitwasobviousfromtheresultsthatparticipantsperformedsignicantlyworsewiththeLEPmotherswhentheywereinterviewedaftertheEPmothers. 75

PAGE 76

REFERENCES [1] Guidancetofederalnancialassistancerecipientsregardingtitleviprohibitionagainstnationalorigindiscriminationaffectinglimitedenglishprocientpersons. [2] Ferri'sdifferentialdiagnosis:,2nded.ed.Elsevier/Mosby,,Philadelphia,PA:,c2011. [3] A,K.,JC,C.,K,J.,andB,L.Simulationofavirtualpatientwithcranialnerveinjuryaugmentsphysician-learnerconcernforpatientsafety.JournalofBio-AlgorithmsandMed-Systems6,11(2010),25. [4] Babu,S.,Suma,E.A.,Barnes,T.,andHodges,L.F.Canimmersivevirtualhumansteachsocialconversationalprotocols?InVR,W.R.Sherman,M.Lin,andA.Steed,Eds.,IEEEComputerSociety(2007),215. [5] Bach,P.B.,Pham,H.H.,Schrag,D.,Tate,R.C.,andHargraves,J.L.Primarycarephysicianswhotreatblacksandwhites.NewEnglandJournalofMedicine351,6(August2004),575.Accesstofulltextissubjecttothepublisher'saccessrestrictions. [6] Bickmore,T.W.,Caruso,L.,Clough-Gorr,K.,andHeeren,T.'it'sjustlikeyoutalktoafriend'relationalagentsforolderadults.Interact.Comput.17,6(Dec.2005),711. [7] BrianD.Smedley,A.Y.S.,AlanR.Nelson,Editors,C.o.U.,Racial,E.,andinHealthCare,E.D.UnequalTreatment:ConfrontingRacialandEthnicDisparitiesinHealthCare(fullprintedversion).TheNationalAcademiesPress,2003. [8] deMelo,C.M.,Carnevale,P.,andGratch,J.Theeffectofexpressionofangerandhappinessincomputeragentsonnegotiationswithhumans.InInternationalConferenceonAutonomousAgentsandMultiagentSystems(AAMAS)(Taipei,Taiwan,May2011). [9] Deaton,E.,Barba,C.,Santarelli,T.,Rosenzweig,L.,Souders,V.,McCollum,C.,Seip,J.,Knerr,W.,andSinger,J.Virtualenvironmentculturaltrainingforoperationalreadiness(vector).VirtualReal.8,3(May2005),156. [10] Decamp,L.,Choi,H.,andDavis,M.Medicalhomedisparitiesforlatinochildrenbyparentallanguageofinterview.JHealthCarePoorUnderserved22,4(2011),1151. [11] Diamond,L.C.,andReuland,D.S.Describingphysicianlanguageuency:deconstructingmedicalspanish.JAMA301,4(2009),426. [12] Diamond,L.C.,Schenker,Y.,Curry,L.,Bradley,E.H.,andFernandez,A.Gettingby:underuseofinterpretersbyresidentphysicians.JGenInternMed24,2(2009),256. 76

PAGE 77

[13] Ferguson,W.J.,andCandib,L.M.Culture,language,andthedoctor-patientrelationship.FamMed34,5(2002),353. [14] Fernandez,A.,Schillinger,D.,Warton,E.M.,Adler,N.,Moffet,H.H.,Schenker,Y.,Salgado,M.V.,Ahmed,A.,andKarter,A.J.Languagebarriers,physician-patientlanguageconcordance,andglycemiccontrolamonginsuredlatinoswithdiabetes:Thediabetesstudyofnortherncalifornia(distance).JGenInternMed26,2(2011),170. [15] Gany,F.,Leng,J.,Shapiro,E.,Abramson,D.,Motola,I.,Shield,D.C.,andChangrani,J.Patientsatisfactionwithdifferentinterpretingmethods:arandomizedcontrolledtrial.JGenInternMed22Suppl2(2007),312. [16] Gratch,J.,Rickel,J.,Andre,E.,Cassell,J.,Petajan,E.,andBadler,N.Creatinginteractivevirtualhumans:Someassemblyrequired.IEEEIntelligentSystems17,4(July2002),54. [17] Jacobs,E.A.,Lauderdale,D.S.,Meltzer,D.,Shorey,J.M.,Levinson,W.,andThisted,R.A.Impactofinterpreterservicesondeliveryofhealthcaretolimited-english-procientpatients.JGenInternMed16,7(2001),468. [18] Johnsen,K.,Dickerson,R.,Raij,A.,Lok,B.,Jackson,J.,Shin,M.,Amy,J.H.,andLind,D.S.Experiencesinusingimmersivevirtualcharacterstoeducatemedicalcommunicationskills.InInProc.IEEEVirtualReality,IEEEPress(2005). [19] Johnsen,K.,Raij,A.,Stevens,A.,Lind,D.S.,andLok,B.Thevalidityofavirtualhumanexperienceforinterpersonalskillseducation.InProceedingsoftheSIGCHIconferenceonHumanfactorsincomputingsystems,CHI'07,ACM(NewYork,NY,USA,2007),1049. [20] Jr.,R.W.H.,Gratch,J.,Marsella,S.,Rickel,J.,Swartout,W.R.,andTraum,D.R.Virtualhumansinthemissionrehearsalexercisesystem.KI17,4(2003),5. [21] Kang,S.-H.,Gratch,J.,Sidner,C.,Artstein,R.,Huang,L.,andMorency,L.-P.Towardsbuildingavirtualcounselor:Modelingnonverbalbehaviorduringintimateself-disclosure.InEleventhInternationalConferenceonAutonomousAgentsandMultiagentSystems(Valencia,Spain,June2012). [22] Karliner,L.S.,Jacobs,E.A.,Chen,A.H.,andMutha,S.Doprofessionalinterpretersimproveclinicalcareforpatientswithlimitedenglishprociency?asystematicreviewoftheliterature.HealthServRes42,2(2007),727. [23] Karliner,L.S.,Perez-Stable,E.J.,andGildengorin,G.Thelanguagedivide.theimportanceoftrainingintheuseofinterpretersforoutpatientpractice.JGenInternMed19,2(2004),175. [24] Kotranza,A.,Johnsen,K.,Cendan,J.,Miller,B.,Lind,D.S.,andLok,B.Virtualmulti-toolsforhandandtool-basedinteractionwithlife-sizevirtualhumanagents.In 77

PAGE 78

Proceedingsofthe2009IEEESymposiumon3DUserInterfaces,3DUI'09,IEEEComputerSociety(Washington,DC,USA,2009),23. [25] Ngo-Metzger,Q.,Sorkin,D.H.,Phillips,R.S.,Greeneld,S.,Massagli,M.P.,Clarridge,B.,andKaplan,S.H.Providinghigh-qualitycareforlimitedenglishprocientpatients:theimportanceoflanguageconcordanceandinterpreteruse.JGenInternMed22Suppl2(2007),324. [26] Pandya,Chhandasi,Batalova,J.,andMcHugh,M.Limitedenglishprocientindividualsintheunitedstates:Number,share,growthandlinguisticdiversity.,2011. [27] Parsons,T.D.,andRizzo,A.Virtualhumanpatientsfortrainingofclinicalinterviewandcommunicationskills.InProceedingsofthe2008InternationalConferenceonDisability,VirtualRealityandAssociatedTechnology(Maia,Portugal,Sept.2008). [28] Pertaub,D.-P.,Slater,M.,andBarker,C.Anexperimentonpublicspeakinganxietyinresponsetothreedifferenttypesofvirtualaudience.Presence11,1(2002),68. [29] Pippins,J.R.,Alegria,M.,andHaas,J.S.Associationbetweenlanguageprociencyandthequalityofprimarycareamonganationalsampleofinsuredlatinos.Medicalcare45(20072007),10201025. [30] Rickel,J.,andJohnson,W.L.Virtualhumansforteamtraininginvirtualreality,1999. [31] Rossen,B.,Johnsen,K.,Deladisma,A.,Lind,S.,andLok,B.Virtualhumanselicitskin-tonebiasconsistentwithreal-worldskin-tonebiases.InIVA,H.Prendinger,J.C.Lester,andM.Ishizuka,Eds.,vol.5208ofLectureNotesinComputerSci-ence,Springer(2008),237. [32] Rossen,B.,Lind,S.,andLok,B.Human-centereddistributedconversationalmodeling:Efcientmodelingofrobustvirtualhumanconversations. [33] Tienda,M.,andFaithMitchell,Editors,C.o.T.O.C.D.N.R.C.HispanicsandtheFutureofAmerica.TheNationalAcademiesPress,2006. [34] Traum,D.,Marsella,S.C.,Gratch,J.,Lee,J.,andHartholt,A.Multi-party,multi-issue,multi-strategynegotiationformulti-modalvirtualagents.InPro-ceedingsofthe8thinternationalconferenceonIntelligentVirtualAgents,IVA'08,Springer-Verlag(Berlin,Heidelberg,2008),117. [35] Wilson,E.,Chen,A.,Grumbach,K.,Wang,F.,andFernandez,A.Effectsoflimitedenglishprociencyandphysicianlanguageonhealthcarecomprehension.JGenInternMed20,9(2005),800. 78

PAGE 79

BIOGRAPHICALSKETCH DoaaElSheikhgraduatedwithaBSdegreeinelectricalengineeringin2004fromCairoUniversity,Egypt.ShereceivedherMSdegreeincomputersciencefromtheUniversityofFloridainfall2012.ShejoinedtheVirtualEnvironmentResearchGroup(VERG)in2010whereshehasbeeninvestigatingthepotentialofvirtualhumansinovercominglanguagebarriersbetweenphysiciansandpatients.ShedevelopedasystemthatallowsmedicalstudentstopracticeperformingaphysicalexamonvirtualpediatricpatientswhoseparentsexhibitdifferentlevelsofEnglishprociency.Herresearchinterestsareinhumancomputerinteractionandvirtualenvironments. 79