Transcription
Real-time Data Warehouse:Oxymoron or Clinical Transformation?Se s s ion # 20 2, Augus t 12, 20 21, 1:15-2:15Cra ig Schwa bl, MBAJe s s e Pre s t on, MBAVice President, Digital Solutions & Enterprise AnalyticsDirector, Enterprise Data ServicesDISCLAIMER: The views and opinions expressed in this presentation are solely those of the author/presenter and do not necessarily represent any policy or position of HIMSS.1
WelcomeCra ig Schwa bl, MBAJe s s e Pre s t on, MBAVice President, Digital Solutions & Enterprise AnalyticsDirector, Enterprise Data Services#HIMSS212
Conflict of InterestCraig Schwabl, MBAHas no real or apparent conflicts of interest to report.Jesse Preston, MBAHas no real or apparent conflicts of interest to report.#HIMSS213
Agenda Background & Challenges Use Cases How we got there Technical Solution Results#HIMSS214
Learning Objectives Explore how critical business decisions can leverage real-time data Discover solutions in overcoming challenges to cloud analytics migration Modify one methodology of cloud migration with real-time data to suit theneeds of your own organization#HIMSS215
Background In 2016, University Hospitals of Cleveland established a formal EnterpriseBusiness Intelligence ProgramBusiness IntelligenceProgram PillarsPortfolioPlatform-Data WarehouseSelf Service ReportingData MiningPredictive AnalyticsMaster Data Mgmt.ETL ProcessesData validation-Data GovernanceStandard “Packages”Business RequirementsRequest validationIntake assessmentPrioritizationKaizen / focus groups-Data DomainsData DefinitionsBusiness owners (akaData Stewards)Data use policiesData access (who)Program Governance-BI StrategyProgram Oversight-Program PrioritizationBusiness Alignment-Program SponsorshipKey Decisions#HIMSS216
Challenges Covid-19 put enormous strain on the Health System Lack of real-time visibility of emerging Covid hotspots Equipment shortagesSupply shortagesStaffing shortagesOutbreaks were occurring quicklyDifficult to predict hospital bed utilizationCovid-19 introduced critical needs for real-time decision-making#HIMSS217
Use Cases Need real-time visibility into ventilator usage & allocation Need real-time geospatial analysis of Covid cases Need real-time allocation/re-allocation of staff Need standards-based real-time integration with partners#HIMSS218
How we got therePhase 1 – Modify current environment for near-real-time data High-frequency Extract, Transform, Load (ETL) processAdvantages Leverage existing data flow pipelines Near real-time data for some data types Straightforward implementationDisadvantages Additional strain on source systems Required hardware upgrades to our Enterprise Data Warehouse (EDW) and ETLenvironments Wasn’t able to satisfy all use cases#HIMSS219
How we got therePhase 2 – Design and implement a real-time, cloud-based HL7/FHIR platform Advantages True real-time data Standards-based Leverages existing messaging infrastructure Scalable cloud-based infrastructureDisadvantages More complex environment Requires new development of HL7 - FHIR mapping Relies on newer, nascent technology (FHIR APIs)#HIMSS2110
Technical SolutionCorporate Data CenterAzure Private rceSystemsFHIRAPIReal-timeData MartSQLData MiningReportingHL7 - FHIRConversionEnterpriseDataWarehouse#HIMSS2111
Technical SolutionHL7 - FHIR MappingAuth, POST - Response to FHIR API*GET from FHIR API*source: fhir.org* All patient names are fictitious#HIMSS2112
Technical SolutionImplementation Considerations: Scalability: Throughput: Need to ensure sufficient bandwidthCode Optimization University Hospitals generates 10M HL7 messages per day ( 300M monthly)Real-time “line speed” mapping to avoid congestion & engine backupPlatform Evolution APIs & mapping files#HIMSS2113
Results“Can monitor the patient’s vital informationminute-by-minute from anywhere on thefloor or even remotely outside the hospital”“With the ability to quickly see all vitalinformation, UH will also be able tomonitor and quickly respond to the needof patients on ventilators”wkyc.com#HIMSS2114
Results Near real-time access to ventilatorusage and utilization Enabled decisions for reallocationof devices across facilities#HIMSS2115
Results Real-time analysis of clusteroutbreaks Predictive analysis of communalliving outbreak risk#HIMSS2116
Results Real-time staffing utilization Predict staffing shortages Reallocation of staff todifferent facilities and/ordepartments#HIMSS2117
Additional Use Cases Support for non-HL7 partner platforms Central operations CensusSchedulingBillingPatient movement Resource Forecasting Extended Capacity Management (e.g. bed availability) Value-based care (mobile) Real-time Exception reporting & alerts Physician Alerting (ED / admissions)Clinical data (e.g. Sepsis temperature) Advanced ETL packages - micro-processing#HIMSS2118
Questions Have any organizations successfully made the leap from on-premise analytics toa cloud environment? If so, what were the business drivers for this transition? For organizations that have migrated to the cloud, what were some of thesuccess and challenges that you faced? For organizations that have implemented real-time data warehouse capabilities,what were the business problems that were being addressed? What source data feeds your real-time data warehouse environment? How is your real-time data being used (e.g. self-service analytics, feed machinelearning models, AI platforms, geo-spatial analysis, data interoperability, etc)?#HIMSS2119
Thank you!(*Reminder to complete the online evaluation of this session)Cra ig Schwa bl, MBAVP, Digital Solutions & Enterprise AnalyticsCraig.Schwabl@uhhospitals.orgJe s s e Pre s t on, MBADirector, Enterprise Data ServicesJesse.Preston@uhhospitals.org#HIMSS2120
Rhapsody Interface Engine. Real-time Data Mart. HL7 Applications. Data Mining. SQL. Source Systems. Azure Private Cloud. FHIR. API. REST. Reporting. Enterprise Data Warehouse. HL7 - FHIR Conversion. #HIMSS21 12 Technical Solution HL7 - FHIR Mapping. source: fhir.org. Auth, POST - Response to FHIR API*
