An Enterprise Data Warehouse Solution That Bridges Best

An Enterprise Data Warehouse Solution That Bridges Best Practices in Data Standards and Quality, Data Governance, and Data-warehousing Operations S 28: Wrangling with Clinical Data for Insights Sarah Oraby, MS, Raj Iyer, MBA Mike Wall, Pharm D, MBA UChicago Medicine, Chicago IL

Disclosure We have no relevant financial relationships with commercial interests to disclose. 2 AMIA 2019 Clinical Informatics Conference

Learning Objectives After participating in this session the learner should be better able to: • To build a scalable infrastructure that can support data growth and hospital system expansion • Develop a data model that is designed to integrate clinical and operational data • Develop a framework that will support first of its kind data standardization through reference data management , metric management and data quality monitoring/anomaly detection • Establish a cross-functional governance framework and processes that will enable the organization to leverage the technology and platform to realize enhanced data output productivity and cost savings 3 AMIA 2019 Clinical Informatics Conference

About UChicago Medicine The University of Chicago Medicine, with a history dating to 1927, is a not-for-profit academic health system based on the campus of the University of Chicago in Hyde Park with hospitals, outpatient clinics and physician practices throughout Chicago and its suburbs. UChicago Medicine unites five organizations to fulfill its tripartite mission of medical education, research and patient care: Biologicial Sciences Division University of Chicago Medical Center UChicago Medicine Physicians Pritzker School of Medicine Community Health & Hospital Division 4 AMIA 2019 Clinical Informatics Conference

UChicago Medicine Hospital System 5 AMIA 2019 Clinical Informatics Conference

An Enterprise Data-Warehouse: The Journey • The problem/gap and the vision • The project team • The product • The organizational shift • Lessons learned and take-aways 6 AMIA 2019 Clinical Informatics Conference

The Driving Force The Problem/ Gap The Vision q Disparate data limits analytics ü Unified/integrated data sources; better and faster analytic outcomes q Uncertain/ no data quality ü Structured/automated data quality management q Redundant preprocessing ü Streamlined operations q Variable data literacy ü Shared decision and knowledge q Lack of governance ü Governance & oversight q Incomplete subject area representation ü Built across all departments/subject areas q Limited infrastructure ü Scalable infrastructure 7 AMIA 2019 Clinical Informatics Conference

The Project Team Project Steering Committee Executive Level Sponsors Program Manager Project Management oversight Subject Area Experts Finance, Clinical, IT, Clarity, Reporting Technical Project Oversight Director-Level Management from Data & Analytics, IT Data Architect (3) Data Staging ETL (2) Programmer (1) Quality Assurance (3) Reference Data Mgmt. (1) Business Analyst (1) ETL Modeler (2) Database Admin (1) 8 AMIA 2019 Clinical Informatics Conference

The Product: A Unified Data Platform (UDP) The Unified Data Platform (UDP) is the preferred UChicago Medicine data and analytics platform that will be used to support clinical and operational initiatives for the health system Why is UDP the preferred platform? UDP uncages data! • Unified: A single platform for all clinical and operational initiatives • Native and Normalized: Ability to deliver data in both a native (source data directly in data lake) and normalized (cleansed and standardized data in EDW) form • Consistent: Supports data consistency through standardized metric definitions and data quality management • Available: Ensures the right data is available to the right users through access control • Governed: Facilitates proper data use and data quality through policies, standards, and guidelines • Efficient: Reduces data pre-processing burden of analysts and end users (e. g. repeatable filters, starter queries) 9 AMIA 2019 Clinical Informatics Conference

3 Main Deliverables of UDP Scalable Infrastructure • Expand servers to include formal dev, test, prod • Ensure scalability to support growth and recovery • Scalable data model to support data landing, data integration, and subject area data marts • Integration of key data sources/subject areas to automate manually generated inbound and outbound regulatory reports Integrated Data Model • Technical framework for data quality, reference data management and metric management • Kick off data governance council and policies • Kick off a data warehouse project prioritization committee Governance • Identify and define dedicated data quality and governance resources and processes 10 AMIA 2019 Clinical Informatics Conference

UDP Capabilities Delivery Roadmap 2018 • Sources Integrated: hand hygiene data, • Implement Proof of Concept for unified architecture and process • Sources Integrated: Clarity, third-party vendor data, cost-accounting data, Patient satisfaction, SS Death Registry, billing data • Extracts: The largest reports: vendor data benchmarking, CMS quality reporting, IDPH regulatory, PHIS • Data Quality framework (with machine learning capabilities) • • • scheduling, patient logistics data, claims, operational data, supply chain Reports: research extract, budget, no show prediction, schedule optimization, all go forward reports Data Quality framework: EDW with applied data quality and data cleansing Expanded subject area data integration • Sources Integrated: Population health, value based care and other external sources, quality data DW, IT datawarehouse have been retired • All analytics and reporting sourced from UDP • Data integration across all subject areas (HR) Sep & Beyond May - Aug Reporting & Analytics Output & Efficiency Integrate additional data sources and onboard early adopters Fully operationalize data platform with proper governance across all functions Jan - April Time Minimal Output & Efficiency 11 High Output & Efficiency AMIA 2019 Clinical Informatics Conference

Governance Tool Governance Presentation UDP architectural vision RDM Analytics and Dash-boarding Tools To Vendors State CMS UDP Oversight DE-ID EDW UDP Huddle EDW ETL OMOP Lake Data Marts DQM Data Lake Big Data Late-Binding Data Quality DQM Reports DQM Editor Meta Data Stand ardizat ion Machine Learning Metric Editor RDM Other’s De-ID Data Automati on DQRules Results Late Binding Metrics Machine Learning Metrics Engine Rules Engine DQM Staging ETL UDP Oversight DQM Native Data Quality Payer Clarity subset Cost Accounting subset Caboodle subset MDM Patient Satisfaction Unstructured Data • ETL • Database Vendor ETL Source Data Governance Council Data Marts Star DW RDM Editor Late-Binding Metrics Payer Clarity Cost Accounting Caboodle MDM Vendor 12 AMIA 2019 Clinical Informatics Conference

The Data Platform consists of: • Data lake: Raw data that is extracted directly from the source • Enterprise data warehouse (EDW): Integrated data that is high quality, fidelity, and value • Data marts: Series of analytics databases specific to the needs of the data consumers (e. g. reports, extracts) 13 AMIA 2019 Clinical Informatics Conference

Governance Structure Data Governance Committee Governance Support Liberate, Cleanse and Apply Demand Management Committee Data Quality Management UChicago Medicine Clinical, Operational, and Administrative Departments IT/UDP Data Science and Analytics 14 DM Program Office Tools, Solutions, policies DG Program Office Guide and Promote Users, Stewards, Experts Steering Committee Oversight and Strategy Data Quality Engineering AMIA 2019 Clinical Informatics Conference

The biggest organizational shifts… 1. The cross functional proof of concept project translated into a crosspollinated data warehousing team with best practices learned from across departments into one new team 2. Redefined governance committees provide a platform for the voices of both data professionals and business leaders where data professionals advocate for data quality and business leaders advocate for prioritization of resources and funding towards the most valuable data pursuits 3. Laying the foundation to support data quality management and related process management with a dedicated team that works with data governance and subject area experts to ensure data integrity of the data warehouse. 15 AMIA 2019 Clinical Informatics Conference

Lessons Learned 1. Prioritizing data requests requires a long hard look at the data that exists, how well its understood and full transparency across the organization leaders. Consider all your data request intake channels. 2. Data Governance council is a decision making body, they cannot review and approve policies and monitor and solve data quality issues. A dedicated team for data quality management must exist to support the data governance council 3. Start governance with education and literacy, organizational changes and process changes around data may result in culture shifts. Minimize confusion and maximize consensus by ensuring the key audiences are well informed of these changes and the roadmap 16 AMIA 2019 Clinical Informatics Conference

Practical Application of this Session • Gives the audience an example project structure and framework for building an enterprise data warehouse platform • Gives the audience critical design elements and types of data that should be included in an enterprise data warehouse platform • Gives the audience critical governance structures, framework and roles that are necessary for supporting an enterprise data warehouse 17 AMIA 2019 Clinical Informatics Conference

AMIA is the professional home for more than 5, 600 informatics professionals, representing frontline clinicians, researchers, public health experts and educators who bring meaning to data, manage information and generate new knowledge across the research and healthcare enterprise. AMIA 2019 Clinical Informatics Conference | amia. org @AMIAInformatics @AMIAinformatics Official Group of AMIA @AMIAInformatics #Why. Informatics 18

Thank you!