Transcription

Annual All-Hands MeetingOctober 13, 2016Progress and work forEffectivenessResearchLucila Ohno-MachadoSupported by the Patient-Centered Outcomes Research Institute Contract CDRN-1306-048191

pSCANNER is a stakeholder-governed,distributed clinical data network that aims to makehealth data more accessible and usable for thegeneration of scientific evidence that patients,clinicians, and other stakeholders together use tomake more informed health decisions.Slide from K Kim and J Doctor2

PCORnet embodies a “community of research” by uniting systems,patients & clinicians13 ClinicalDataResearchNetworks(CDRNs)20 PatientPoweredResearchNetworks(PPRNs)PCORnet:A nationalinfrastructure forpatient-centeredclinical research3

Collaborations across the countryThis map depictsthe number ofPCORI-fundedPatient-Powered orClinical DataResearch Networksthat have coveragein each state.Numbers donot include theVA4

NetworkCTSA hubPhase 124 MillionPatients9 healthsystemsUC ReXUCD2.3MVA VINCI 11MUCSF3.2MUCLA4.3MUCI300kUCSD2.3MUSC LAAltaMedTCCQueensCare607k240k19kSCANNER5

NetworkCTSA hubUniversity of Washington CTSAPhase 230 MillionpeopleUC ReXUniversity of ColoradoUC Davis2.3MSan MateoMedicalCenter 77kVA VINCI 11MLAUCSF3.2M14 healthsystemsUSC Keck 2MUCLA4.3MCedarsSinai 2MUC Irvine300kUCSD2.3MLA Children’s 200kLA DHS 600KAltamed 200kChidlren’s Clinic 24kQueenscare 19kRutgersEmory USCANNERU Texas Houston6

Collaborating Patient-PoweredResearch NetworksPPRNTargeted Condition (IRB approved)Health eHeartHeart DiseaseAR-PoWERInflammatory ArthritisCENAAlström Syndrome, Dyskeratosis Congenita, Gaucher Disease,Hepatitis, Inflammatory Breast Cancer, Joubert Syndrome, KlinefelterSyndrome and Associated Chromosomal Anomalies, MetachromaticLeukodystrophy, Pseudoxanthoma ElasticumDuchenneConnectDuchenne and Becker Muscular DystrophyiConquerMSMultiple SclerosisPPRN (in process)Targeted ConditionImproveCareNowInflammatory Bowel DiseasePRIDEnetSexual and Gender MinoritiesCrohn's & Colitis Foundation ofAmerica (CCFA)Crohn's Disease and Ulcerative ColitisMoodNetworkMood DisordersSlide from Dr. Maehara7

Studies are already underway, at the same time we are stillrunning queries to assess data quality ADAPTABLE: clinical trial on aspiring dosage (UCLA) Bariatric surgery outcomes (UCI) Diabetes Risk, eCQM (USC) COPD medications8

Harnessing PCORnet to Study ComparativeEffectiveness and Safety of Biologic TherapiesPI: Jeffrey Curtis, UABPPRN Research Demonstration Project (2016-19)Immune-mediated diseases affect over 8 million patientsRapidly changing treatment landscape, with effective, but expensive,immunosuppressive and immunomodulatory therapies, with concerningsafety profile (cancer, serious infections, cardiovascular events)Aim #1: To evaluate comparative risks for serious infections,cardiovascular risk, incident cancer and mortality, with biologics and nonbiologics for adult and pediatric immune-mediated diseasesAim #2: To evaluate the comparative clinical effectiveness of variousmedications using patient-generated data for a variety of disease-specificand generic patient reported outcome (PRO) domains9

A Network of NetworkspSCANNER will enable researchers to use data from distributed sites coveringover 30 million patients in a privacy-preserving environment.VA11MCedars Sinai2MUCSFUCD3M2M.2MCHLAUSC.3MALTAMED2MpSCANNER Hub4MUCLA.2MTCC.2M.8M.3M2MSan MateoUCIUCSDQUEENSCARE10

Preserving Privacy11

Sharing Data, Tools, SystemsData sharingecosystemdifferential privacyindexinghomomorphic encryptionsecure multiparty computationOhno-Machado L. To Share or Not To Share: That Is Not the Question. Science Translational Medicine, 2012 4(165)12

Distributed architectureAnalysis: Distributed computing Predictive modeling and adjustment for cofounders require lots of data Some institutions (countries) cannot move data outside their firewallsUser requests data forQuality Improvement orResearchDiverse Healthcare Entitiesin 3 different states (federal, state, private)TrustedBroker(s) Identity & TrustManagement PolicyenforcementSecurity EntityWu Y et al. Grid Binary LOgistic REgression (GLORE): Building Shared Models Without Sharing Data. JAMIA, 2012Wang S et al. EXpectation Propagation LOgistic REgRession (EXPLORER): Distributed Privacy-Preserving Online Model Learning. J Biomed Inf 2013Jiang W et al. WebGLORE: A Webservice for Grid Logistic Regression. Bioinformatics 2014Wu Y et al. Grid Multi-Category Response Logistic Models. BMC Med Inform Dec Making 201513

Horizontal and Vertical nsuranceB145YB232YPatientAgeInsuranceA145XA232YLi Y, Jiang X, Wang S, Xiong L, OhnoMachado L. VERTIcal Grid lOgisticregression (VERTIGO) –J Am Med Inf Assoc. 201614

Related Work in Patient-control of DataPatient ionsDo I wish todisclose data Dto U?Trusted broker Data useagreementsPatient ISharingLook-up Study registryUser U requestsData D onindividual II can checkthat U lookedat my data D15

iCONCURInformed CONsent forClinical data Use inResearch1616

Courtesy of HKimNIH R01HG0088021717

iCONCUR Asked 1kpatients 394 wereinterested in thestudy 124 went to website to makeselections 20% Cal. PMINIHR01HG008802 70% National PMIKim H et al, JAMIA201618

Stay tuned for today’spresentations(and re-watch them online too athttp://pscanner.ucsd.edu)Supported by the Patient-Centered Outcomes Researach Institute Contract CDRN-1306-0481919

ADAPTABLE: clinical trial on aspiring dosage (UCLA) Bariatric surgery outcomes (UCI) Diabetes Risk, eCQM (USC) COPD medications . 9 Immune-mediated diseases affect over 8 million patients Rapidly cha