An Informatics Architecture for an Exposome v1...
Transcript of An Informatics Architecture for an Exposome v1...
DepartmentofBiomedicalInformatics
BiomedicalInformaticsCoreCenterforClinicalandTranslationalSciences
AnInformaticsArchitectureforanExposome
SessionII06– SecondaryUseofDataforResearch(InteractiveLearning)AMIA2016JointSummitsonTranslationalScience
March22nd,2016
Dr.RamGouripeddiAssistantProfessor,DepartmentofBiomedicalInformaticsChiefBiomedicalInformaticist,BiomedicalInformaticsCore,
CenterforClinicalandTranslationalScienceUniversityofUtah
Acknowledgements
• PRISMSPI(s)• KatherineSward,RN,PhD• JulioC.Facelli,PhD
• PRISMSTeamatUniversityofUtah• Nodisclosures
This effort was supported by U54EB021973, National Institute of Biomedical Imaging and Bioengineering, NIH andUniversity of Utah Air Quality Seed Grant Program. OpenFurther supported by NCRR/ NCATS Grants UL1RR025764 and3UL1RR025764-02S2, National Center for Clinical and Translational Science 1UL1TR001067, University of Utah ResearchFoundation, grant 1D1BRH20425 (DHHS), and R01 HS019862 from AHRQ, (DHHS).
2
Overview
• EffectsofEnvironmentonHealth• KeyConcepts• InitialWork– AMIA2014• Limitations• ChallengesandInformaticsMethodsandSolutions• PRISMS• InformaticsArchitecture
3
EffectsofEnvironmentonHealth• Phenotype:Resultofinteractionsbetweengenotypeandenvironment.• Environmentalfactorscontributesignificantlybythemselvesandtheirinteraction1 withbehavioral,occupationalandmetabolicfactors1.
Disability-adjusted life-years attributable to behavioral,environmental, occupational, and metabolic risk factors1.4
FlintWaterCrisis• LeadPoisoninginkids
• Immunedisorders• Criminaltendencies• Behaviorandlearningproblems• LowerIQandhyperactivity• Slowedgrowth• Hearingproblems• Anemia
• Noknownsafelevelofleadinachild'sblood.• LeadActionLevel:10%ofdrinkingwater>10partsperbillion.• CDC’spublichealthactions:whenthelevelofleadinachild’sblood≥5microgramsperdeciliter2.
http://electrochemistryresources.com/wp-content/uploads/2016/02/corrosion-water-pipe.jpg
5
Exposome3-6
• Encompasseslife-courseofenvironmentalexposures(includinglifestylefactors)fromprenatalperiodonwards.• Complementsgenomebyprovidingacomprehensivedescriptionoflifelongexposurehistory.
GeneralExternalEnvironment
SpecificExternal
EnvironmentInternalEnvironment
Overlappingdomainswithinexposome6
Exposomics
• Studyofdefining,generatingandutilizingexposomes inbiomedicalresearch.• Ongoingefforts:
• HELIX7:Earlylifeexposome• EXPOsOMICS8:Assessexposures• HEALS9:Studiesexposuretoenvironmentalstressorsandhealthoutcomes• NIH’sEnvironmentalinfluencesonChildHealthOutcomes(ECHO)Program10:Understandingtheeffectsofenvironmentalexposuresonchildhealthanddevelopment
• Requiresasystemsbiologyapproach.• ‘Expotying’:Exposureofabiologicalentityusuallywithreferencetoaspecificcharacteristicunderconsideration.• AlsocalledasExposomeInformatics,ExposureInformationScience.• ProvidesgreatopportunitiestoBiomedicalInformatics11.
7
DefiningandGeneratinganAirQualityExposome
8
Background• AirQuality(AQ)hasbeenassociatedwithvariousadversehealtheffects
• Asthma• Cardiovasculardisease• Respiratoryinfections• Cancers• Impairedglucosetoleranceduringpregnancies12-15.
• ResearchersattheUniversityofUtahareembarkingonclinicalstudiestounderstandassociationsbetweenthepeculiarAQpatternsinSaltLakeCityandclinicalconditions:• Cerebralvenousthrombosis• Exacerbationsofidiopathicpulmonaryfibrosis• Suicide• Reproductiveoutcomes• Cancers.
9
SaltLakeCityAirQuality
• Pronetowinterinversionswherecoldersurfacetemperaturestrapfineparticulatematter(PM2.5)whichposesserioushealthconcerns.• Summermonthsinthevalleyhaveincreasedozone(O3)levels16.• Natural/Quasi-experimentalconditions.
Courtesy:Dr.K.Kelly
10
OpenFurther17-18
• QueryTool• FederatedQueryEngine
• DataSourceAdapters• Admin&SecurityComponents
• VirtualIdentityResolutionontheGO(VIRGO)
• Quality&AnalyticsFramework
• MetadataRepository• Terminology/OntologyServer
• AirQualityModellingUnit
11
OpenFurtherDeploymentsandUses
CohortSelection,UniversityofUtah
ComparativeEffectivenessResearch,
PHIS+
CohortSelection,UniversityofNorth
Carolina
DataIntegration&AnalyticsPipeline,UtahDepartmentofHealth
12
AirQuality- ClinicalDataFederation
• DemonstratedfeasibilityoffederatingairqualitydatafromEnvironmentalProtectionAgency(EPA)withclinicaldatafromUniversityofUtahusingOpenFurther19-20.• AbilitytoselectdifferentcohortsofpatientslivinginSLCcountyandhavingclinicalconditions(e.g.asthma)occurrencesthatwererelatedtotemporalvariationsofairpollutantconcentration.
13
AirQualityMonitoringinSaltLakeCounty
• ThreemonitoringstationsinSaltLakeCounty.• AQspeciesconcentrationvariationsduetotopography,altitudeandmeteorology21-22
• Whatistheairqualityatanyotherlocation?• Needforcross-linkingpatientlocationsandconditionoccurrences:HighResolutionSpatio-temporalAirQualityGrid
14
AirQualityExposome
Pollutant&Quantity
Travel Home Ventilation
Outdoors ClinicalConditions
BiologicalMembranes
Others
AirQuality Socio-economic Behavioral Clinical/Physiological Genomic Proteomic
15
BiomedicalResearchAirQualityRequirements
• Primaryneed:understandrisksassociatedwithbeingexposedwithvariousairpollutants.• Manifestationsfollowingexposurecouldoccur• Immediately• Afteralagphase• Couldpersistoverlongdurations.
• Needforunderstandingpathophysiologyandmechanismsofthesemanifestations.• Currentresearchmainlyassociatessinglepollutantandclinicalconditions,futureareasofresearchcouldincludeexposurestomultiplepollutants.
16
UtilizingAirQualityDatainBiomedicalResearch
• IntegratingAQandbiomedicaldataneedstosupport• Spatio-temporalvariationsofairpollutantspecies.• Heterogeneousdata.• Locationofindividuals.• Timingoftheoccurrenceofevents.
• AQdataandresearchrequirementgranularitiesvaryfrominstantaneoustolongerdurationaveragesdepending.• SimplificationofunderstandingandintegratingAQdatawithbiomedicaldata.• Supportbench,translational,clinicalandpopulationresearch.
17
ChallengesandInformaticsMethodsandSolutions
DataSources
MathematicalModeling
UncertaintyCharacterization
DataIntegration• Semantics• Metadata• Time&eventmodeling• Infrastructureformulti-scale,multi-omicsintegration
Presentation/Visualization• Salientfeatureextraction
18
UniversityofUtah’sPRISMSInformaticsInfrastructure
19
PediatricResearchusingIntegratedSensorMonitoringSystems(PRISMS)• Sensor-based,integratedhealthmonitoringsystemsformeasuringenvironmental,physiological,andbehavioralfactorsinpediatricepidemiologicalstudiesofasthma,andeventuallyotherchronicdiseases23.• SensorDevelopmentProjects• InformaticsPlatformTechnologies• DataandSoftwareCoordinationandIntegrationCenter
• UtahTeam:ElectricEngineering,ChemicalEngineering,ComputerScience,AtmosphericSciences,IndustrialEngineering,Informatics,SoftwareDevelopers,Nursing,Pediatrics.
20
AirQualityDataSources
Differentairqualityspecies• ParticularMatter:PM2.5,PM10,UPF• Ozone• CarbonMonoxide• NOx (nitricoxideandnitrogendioxide)• Sulphur Dioxide• Lead• WaterVapor• CarbonDioxide• VolatileOrganicCompounds
Choiceofselectablesourcesforeachspecies
Highresolutionspatio-temporalAQgrid• Personalization
21
TypesofAirQualitySources
PersonalSensors LaserCeilometers NovelSensors MobileSensors
BalloonSensors Satellite-derivedaerosolopticaldepth
measurements
StateEnvironmentalDepartmentNetworks
EnvironmentalProtectionAgency
22
AirQualityMathematicalModels
• Validatedontheeastcoast• Doesn’tconsiderAltitude• 12kilometerresolution• HierarchicalBayesianmodel
EnvironmentalProtectionAgency–CenterforDiseaseControlModel24
• Describeregionalandsmall-scalespatialandtemporalgradients• UsesmeasuredPMconcentrations,monitoringsitelocation,GIS-basedlocation-specificcharacteristicsandlocation-andmonth-specificmeteorologicaldata,andspatialsmoothingofmonthlyandlong-termaverages
GeneralizedAdditiveMixedModels25
• Fillgapsinmeasureddatawithmathematicalmodels.• AlibraryofAQdatamodelstoprovidehighspatio-temporalresolutionwithaframeworkvalidatethemodeloutput.
23
UncertaintyCharacterization26
• SelectionofappropriateAQsourcesandmodels• Inherent:Variationsinunknownconditions• Reducible:Associatedwiththemodelandinputconditions.• ExposureUncertainty:Arisingduetodifferencesinperson’sexposureandtrueambientAQlevels. 24
OpenFurtherModifications1 2 3
45
25
SemanticsforDataIntegration• SemanticinteroperabilityforInternetofThings(IoT)27
• StoredinTerminology/OntologyServer• Examples
• Describessensorsandobservations,andrelatedconcepts.SemanticSensorNetworkOntology28
• StandardmodelsandXMLschemafordescribingsensorssystemsandprocessesassociatedwithsensorobservations.
SensorModelLanguage(SensorML)29
• Standardmeasuresrelatedtocomplexdiseases,phenotypictraitsandenvironmentalexposures.PhenX PhenotypicTerms30
• Facilitatecentralizationandintegrationofexposuredatatoinformunderstandingofenvironmentalhealth.Exposureontology(ExO)31
• GeneOntology,UniProt,SNOMEDetc.Standardbiomedicalontologiesandterminologies 26
Metadata
• StoredinMetadataRepository18
• Relationalorgraphstores• Stores• SourceandCentralDataModels
• Harmonizedsensordatamodel• Dataprovenanceandassociateduncertainty• Inter-modeltransformativefunctions
27
Time&Events
• Datamodeledandstoredinprimitiveformonatimelineasevents.• Transformedtohigher/analyticalmodelsbasedonuse-cases.• Timemodeledas32:• Unbounded:Containsupperand/orlowerboundswithrespecttoitsorderrelationship.• Dense:aninfinitesetofsmallerunits.• Discrete:everyelementhasbothanimmediatesuccessorandanimmediatepredecessor,ifunbounded,andwithintheboundsifbounded.• Instants&Intervals(upperandlowertimepoints).• Finestgranularityavailablewiththesource.
28
DataIntegrationWorkflow
1. Usercanqueryforacohortorcompletedatasets.2. (a&b)Heterogeneousdatasources(whereAandB
representmobile sensordatasources,CrepresentenvironmentalmonitoringdatasourcesandDrepresentbiomedical datasources),and(ifneeded)mathematicalmodelsusingEDMUareselected.• Environmentaldata(A,B&C)harmonizedtothecentral
modelsstoredinMDR.Selectionofmathematical modelsmanagedintheEDMU.
3. OFsynthesizeresultsindifferentanalytical models.4. Presentsthemascohortsand/oraggregatedresults. 29
ResearchUse-Cases
Retrospective&Population
Studies
30
Conclusion
• Scalableinformaticsarchitecturethatisgeneralizablebeyondairqualityandpediatricasthma.• Integratesmulti-scaleandmulti-omicsdata.• Genome-phenome-exposome
• BigDataintegration:volume,velocity,variety,veracityforresearchvalue.• Robustpipelineforresearchdatadeliverywithdecisionsupport.• Supportdifferenttypesofresearch.
31
References1. J.N.Newton,A.D.Briggs,C.J.Murray,D.Dicker,K.J.Foreman,H.Wang,M.Naghavi,M.H.Forouzanfar,S.L.Ohno,R.M.
Barber,andothers,“ChangesinhealthinEngland,withanalysisbyEnglishregionsandareasofdeprivation,1990–2013:asystematicanalysisfortheGlobalBurdenofDiseaseStudy2013,”TheLancet,vol.386,no.10010,pp.2257–2274,2015.
2. USEPA,“BasicInformationaboutLeadinDrinkingWater.”Available:https://www.epa.gov/your-drinking-water/basic-information-about-lead-drinking-water.
3. C.P.Wild,“Complementing theGenomewithan‘Exposome’:TheOutstandingChallenge ofEnvironmentalExposureMeasurementinMolecularEpidemiology,”CancerEpidemiol BiomarkersPrev,vol.14,no.8,pp.1847–1850,Aug.2005.
4. C.P.Wild,“Theexposome: fromconcepttoutility,”Int.J.Epidemiol.,vol.41,no.1,pp.24–32,Feb.2012.5. G.W.MillerandD.P.Jones,“TheNatureofNurture:RefiningtheDefinitionoftheExposome,”Toxicol.Sci.,vol.137,no.1,pp.
1–2,Jan.2014.6. G.W.Miller,TheExposome:APrimer,1edition.Amsterdam ;Boston:AcademicPress,2013.7. “HEALS,”HEALS.Available:http://www.heals-eu.eu/.8. EXPOsOMICS.http://www.exposomicsproject.eu/9. “Home- HELIX|Buildingtheearlylifeexposome.”Available:http://www.projecthelix.eu/.10. NIH’sEnvironmentalinfluencesonChildHealthOutcomes(ECHO)Program.https://www.nih.gov/echo11. F.MartinSanchez,K.Gray,R.Bellazzi, andG.Lopez-Campos,“Exposome informatics:considerations forthedesignoffuture
biomedical research informationsystems,”JournaloftheAmericanMedicalInformaticsAssociation,vol.21,no.3,pp.386–390,May2014.
12. Kesten S,Szalai J,Dzyngel B.Airqualityandthefrequencyofemergencyroomvisitsforasthma.AnnAllergyAsthmaImmunolOffPubl AmColl AllergyAsthmaImmunol.1995Mar;74(3):269–73.
13. Weisel CP,ZhangJ,TurpinBJ,etal.RelationshipsofIndoor,Outdoor,andPersonalAir(RIOPA).PartI.Collection methodsanddescriptiveanalyses.ResRepHealthEffInst.2005Nov;(130Pt1):1–107;discussion109–127.
14. Fleisch AF,GoldDR,Rifas-Shiman SL,etal.AirPollutionExposureandAbnormalGlucoseToleranceduringPregnancy:TheProjectVivaCohort.EnvironHealthPerspect.2014Feb7[cited2014Mar7];http://ehp.niehs.nih.gov/1307065/
15. Kloog I,Koutrakis P,Coull BA,LeeHJ,SchwartzJ.AssessingtemporallyandspatiallyresolvedPM2.5exposures forepidemiological studiesusingsatelliteaerosolopticaldepthmeasurements.AtmosEnviron.2011Nov;45(35):6267–75.
16. UtahConcludesWinter InversionSeason,ResidentsProactivelyEngaged.http://www.deq.utah.gov/News/docs/2014/03Mar/DAQ_NewRelease_AirQualityStats_draftv2.pdf 33
References17. Openfurther.org18. GouripeddiR,SchultzND,BradshawRL,etal.FURTHeR:AnInfrastructureforClinical,TranslationalandComparative
EffectivenessResearch.AmericanMedical InformaticsAssociation,2013AnnualSymposium;2013Nov16;Washington,D.C.http://knowledge.amia.org/amia-55142-a2013e-1.580047/t-10-1.581994/f-010-1.581995/a-184-1.582011/ap-247-1.582014
19. Gouripeddi,R.,andJulioCFacelli. “ProgrammaticallyLinkingAirQuality IndicatorswithClinical Data.”presentedattheAirQuality,PeopleandHealth,2ndAnnualRetreat,UniversityofUtahGuestHouse,UniversityofUtah,SaltLakeCity,April14,2014.http://www.airquality.utah.edu/files/2014/04/Ram_Programmatically-Linking-Air-Quality-Indicator-with-Clinical-Data.pdf.
20. Gouripeddi,R.,Rajan,N.S.,Madsen,R.Warner,P.B.,Facelli, J.C.,FederatingAirQualityDatawithClinical Data,AnnualSymposiumoftheAmericanMedical InformaticsAssociation,2014
21. WhitemanCD,HochSW,Horel JD,Charland A.Relationshipbetweenparticulateairpollutionandmeteorological variablesinUtah’sSaltLakeValley.Atmos Environ.2014Sep;94:742–53.
22. G.D.Silcox,K.E.Kelly,E.T.Crosman,C.D.Whiteman,andB.L.Allen,“Wintertime PM2.5concentrationsduringpersistent,multi-daycold-airpoolsinamountainvalley,”AtmosphericEnvironment,vol.46,pp.17–24,Jan.2012.
23. PediatricResearchUsingIntegratedSensorMonitoringSystems.https://www.nibib.nih.gov/research-funding/prisms24. McMillan,NancyJ.,DavidM.Holland,MicheleMorara,andJingyu Feng.“CombiningNumericalModelOutputandParticulate
DataUsingBayesianSpace–timeModeling.”Environmetrics 21,no.1(February1,2010):48–65.doi:10.1002/env.984.25. Yanosky,JeffD.,ChristopherJ.Paciorek,FrancineLaden,JaimeE.Hart,RobinC.Puett,Duanping Liao,andHelenH.Suh.
“Spatio-TemporalModelingofParticulateAirPollutionintheConterminousUnitedStatesUsingGeographicandMeteorological Predictors.”EnvironmentalHealth13,no.1(August5,2014):63.doi:10.1186/1476-069X-13-63.
26. Burnett,N,Mo,P,Rajan,NS,Madsen,R,Gouripeddi,R.Facelli, JC,AFrameworkforValidatingModeledAirQualityDataforuseinBiomedicalResearch,EnvironmentandSustainabilityResearchSymposium,February17,2015,UnionBallroom,UniversityofUtah,SaltLakeCity.Utah.
27. ALLIANCEFORINTERNETOFTHINGSINNOVATION,“SemanticInteroperability,”2015.28. SemanticSensorNetworkOntology:https://www.w3.org/2005/Incubator/ssn/ssnx/ssn29. SensorModelLanguage(SensorML):http://www.opengeospatial.org/standards/sensorml30. PhenX PhenotypicTerms:https://bioportal.bioontology.org/ontologies/PHENX31. Exposureontology(ExO):https://bioportal.bioontology.org/ontologies/EXO32. C.Combi,E.Keravnou-Papailiou,andY.Shahar,TemporalInformationSystemsinMedicine,2010edition.NewYork:Springer,
2010. 34