Strategies for Achieving Data Liquidity At Duke Eric

Strategies for Achieving Data Liquidity At Duke Eric Poon, MD MPH FACMI Professor of Medicine, Duke University School of Medicine Chief Health Information Officer, Duke University Health System 1

Paradigm Experience? Funding? Research Lifecycle? Ideation Prof Problem & Question Associate Prof Assistant Prof Medical Instructor GME Adequate Funding Lever Identification Intervention Design No Funding Implementation Data Gathering Evaluation 2/19/2021 Publish & Scale-up 2

“I Need EHR Data” 2/19/2021 3

“I Need EHR Data” “…on a shoestring budget” 2/19/2021 4

Self-Service: Data Discovery Tools • Self-service tools for access to EHR data – DEDUCE – Slicer. Dicer – I 2 b 2: access data from the Carolina’s Collaborative ( UNC-Chapel, Wake Forest, and Health Sciences of South Carolina) • Pre-Built Subject Matter: Datamarts, Universes, and Cubes, Dashboards, and Reports 2/19/2021 5

Self-Service: Data Science Pipeline Application Programming Interface (API) 2/19/2021 7

Enterprise Grade APIs Commit Build Test Deploy Cloud Ready Git commit Continuous Integration Version Control Automated Testing Reusable Epic EHR Data Sources Containerization Continuous Deployment Countless Possibilitie s! Distribution of work Scalability Authorization Auditing Clinical Notes Radiology Reports Radiology Image Pathology Reports My. Chart Messages 8

Data Architecture for Now & Future 9

“I Need EHR Data” “…and I have some funding” 2/19/2021 10

Fee-for-Service: Custom Data Set Preparation • Fee-for-Service – Funded through grants or departmental funding – Fee is based upon a flat hourly rate – Opportunities to be linked with Statistical Resources the Biostatistics, Epidemiology, and Research Design (BERD) Methods Core 2/19/2021 11

Fee-for-Service: Custom Data Set Preparation • Fee-for-Service – Consultative service to provide the best data delivery method to meet the need – Tailored combinations of EHR, administrative, patient-provided, and research data – Snapshot research databases made available to PACE or approved locations 2/19/2021 12

“I Need EHR Data” “…and I have sufficient of funding” 2/19/2021 13

Dedicated Analysts and/or Teams • Exclusive Focus on Customers’ Priorities – Supported by Central Analytics Team – One or more Full Time Effort (FTE) dedicated to a department, institute, or initiative – Projects are generally performed in an enterprise fashion with dedicated management – Direction can come from single appointed individual or established governance 2/19/2021 14

Duke Data Prospector Program • Application & Approval Process • Access & Auditing • Training Program • Resources 2/19/2021 15

Data Prospector Program – Curriculum & Support • Standard Epic Training o Epic Clarity o Epic Caboodle • Duke-Specific Curriculum • Dedicated Support o Central Analytics Team Residency (initial) o Ongoing Buddy o Documented Use-cases 2/19/2021 16

Where Do I Start? Gather Data ? QI vs Research ? Consult ? Funded? App Ideas Stage External Collab 2/19/2021 17

EHR Enabled Research Support Group Investigators can request a free consultation to review the investigator's request in order to: • Define a technical approach for a project timeline • Connect the investigator with Duke resources • Refine budget estimates CTSI DOCR CDRN MC Research Forge EHR Enabled Research Support DIHI ACE FFS DCRI Get. Data@Duke ORI 2/19/2021 Biostats Core Mobile App Gateway MC Clinical Systems 18

Acknowledgements • Armando Bedoya, MD MMCi, Associate CMIO, Analytics & Applied Data Science • Stephen Blackwelder, Ph. D, Chief Analytics Officer • Michael Pencina, Ph. D, Vice Dean for Technology and Data Science • Ryan Craig, Senior Director, Enterprise Analytics • Mary Schilder, RN BSN MA, Senior Director, Analytics Customer Solutions 2/19/2021 19

Thank You! Eric. Poon@duke. edu @Dr. Eric. Poon 2/19/2021 20