Standardizing EHR Data for Bridge2AI

Bridge2AI Standards Module

Kyle Zollo-Venecek

Software Development Analyst II

Tufts Clinical and Translational Science Institute (CTSI)

Office Hours 11 May 2023

Overview

• What is OHDSI? What is OMOP?
• What to expect
• From an OMOP ETL
• Expectations of contributing sites
• Support from Standards Module

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

2

Overview (cont)

• Standards Module Curated Resources
• OHDSI ETL Resources
• Source-specific Resources
• How to work through an OMOP ETL

​

​

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

Personal Introductions

• What is your institution?
• What is your role?
• What is your level of familiarity with OMOP/OHDSI?

​

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

OHDSI

• Observational Health Data Sciences and Informatics program
• Global community of researchers and healthcare professionals dedicated to improving health outcomes through the use of standardized, real-world data.
• Has developed several tools and resources for the analysis of observational data, including the OMOP CDM (Observational Medical Outcomes Partnership Common Data Model)
• https://www.ohdsi.org/

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

OMOP Common Data Model

• Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM)
• The OMOP is a public–private partnership that is an initiative of the Foundation for the National Institutes of Health, a 501(c)(3) organization.
• Focused on human care data
• Patient Centric Design
• Worldwide community supported
• 453 data sources: 374 EHRs, 34 registries, 30 administrative claims
• 41 countries
• Estimated 12% global population (~928M) has representation in an OMOP dataset
• 475 publications from the OHDSI community (44% clinical focus)

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

6

OMOP Common Data Model�

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

7

OMOP Version 5.4

https://ohdsi.github.io/CommonDataModel/index.html

What is the scale of an OMOP ETL?

• The scale of an OMOP ETL can vary depending on several factors, such as number of data sources, size of data sets, complexity of source data models, and project goals

​

• Factors to consider when assessing scale include source data prioritization, standardization, technical and domain expertise, and infrastructural availability

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

What CHoRUS expects from you

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

What CHoRUS expects from you

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

What CHoRUS expects from you

• Use resources provided to learn more about OMOP ETLs
• Develop an ETL from your source data to the OMOP CDM
• Run OHDSI software packages on OMOP CDM to derive characterization and data quality reports
• Prioritize Critical Care EHR flowsheet data for mapping and OMOP CDM integration

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

What CHoRUS expects from you

• Use resources provided to learn more about OMOP ETLs
• Develop an ETL from your source data to the OMOP CDM
• Run OHDSI software packages on OMOP CDM to derive characterization and data quality reports
• Prioritize Critical Care EHR flowsheet data for mapping and OMOP CDM integration

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

What help you can expect from us

• Design and implement a centralized ETL support system to assist data contributing sites with their ETL to OMOP CDM
• Provide guidance to sites on the use of OHDSI tools and the OMOP Vocabulary v5.0 in the ETL process
• Providing feedback for sites’ ETL and code mappings
• Offer detailed instructions to sites on how to create a comprehensive OMOP ETL specification
• Contribute to the development of the OMOP Vocabulary by identifying and addressing gaps in both its content and structure

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

Standards Module Resources

• Sample ETLs
• https://drive.google.com/drive/folders/1A212ldq685Milyh07DjfxRd-Ayt0LZ3B

​

• Data Prioritization
• https://drive.google.com/drive/folders/1AaGYoiHDPN6xITxilnEhWNb9Q4181aaI

​

• Office Hours (slides and recordings)
• https://drive.google.com/drive/folders/1ZIeB4blopefJEKp5_m1Y0hHIl4NrEr8g

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

CHoRUS Standards Module Google Drive (more resources will be added):

https://drive.google.com/drive/folders/1hK9SE3VH6Wei7s6HtbtK5s4PJugyrHnD

OHDSI ETL Resources

• The Book of OHDSI
• https://ohdsi.github.io/TheBookOfOhdsi/
• OMOP CDM
• https://ohdsi.github.io/CommonDataModel/cdm54.html
• OMOP ETL Best Practices
• https://www.ohdsi.org/web/wiki/doku.php?id=documentation:etl_best_practices
• OHDSI Forums
• https://forums.ohdsi.org/
• OHDSI YouTube
• https://www.youtube.com/@OHDSI

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

OHDSI ETL Resources (cont)

• EHDEN Academy
• https://academy.ehden.eu/
• Perseus OHDSI Symposium Showcase
• https://www.ohdsi.org/2021-global-symposium-showcase-79/
• 10-Minute Tutorials: ACHILLES
• https://www.youtube.com/watch?v=UyS-LAUql-A
• Running ACHILLES on Your CDM
• https://ohdsi.github.io/Achilles/articles/RunningAchilles.html
• Data Quality Dashboard Tutorial
• https://www.youtube.com/watch?v=RSUgYA6_Kb4

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

Source-specific Resources

• Epic
• Epic UserWeb
• https://userweb.epic.com/
• Clarity Dictionary
• https://datahandbook.epic.com/ClarityDictionary
• Epic Hyperspace: Data Dictionary
• Located in the Analytics Catalog of the Hyperspace Application
• Cerner
• uCERN
• https://community.cerner.com/

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

How to work through the OMOP ETL

• How to get started
• “I really don’t know much about the OMOP Common Data Model”
• The Book of OHDSI (https://ohdsi.github.io/TheBookOfOhdsi/)
• Chapter 2 Where to Begin (https://ohdsi.github.io/TheBookOfOhdsi/WhereToBegin.html)
• Chapter 4 The Common Data Model (https://ohdsi.github.io/TheBookOfOhdsi/CommonDataModel.html)
• Chapter 5 Standardized Vocabularies (https://ohdsi.github.io/TheBookOfOhdsi/StandardizedVocabularies.html)
• Chapter 6 Extract Transform and Load (https://ohdsi.github.io/TheBookOfOhdsi/ExtractTransformLoad.html)
• EHDEN Academy (https://academy.ehden.eu/)
• OMOP CDM and Standardised Vocabularies (https://academy.ehden.eu/course/view.php?id=4)
• Extract, Transform and Load (https://academy.ehden.eu/enrol/index.php?id=7)

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

How to work through the OMOP ETL

• What to do with general questions?
• “Which columns comprise the person table?”
• OMOP Common Data Model (https://ohdsi.github.io/CommonDataModel/index.html)
• OMOP CDM Specification Version 5.4 (https://ohdsi.github.io/CommonDataModel/cdm54.html)

​

• “How do I calculate quantity in the drug_exposure table?”
• OHDSI Forums (https://forums.ohdsi.org/)
• Start with a search, then ask your question to the community

​

• “Which Epic Clarity tables and columns can I use to populate the observation table”
• Epic UserWeb (https://userweb.epic.com/)

​

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

How to work through the OMOP ETL

• What to do with B2AI/CHoRUS-specific questions?
• Contact the Standards Module team via email
• Marty Alvarez (marta.alvarez@tuftsmedicine.org)
• Andrew Williams (andrew.e.williams@tuftsmedicine.org)

​

​

© Tufts Medicine 2023 | Private and confidential. Not for redistribution.

Thank You

Kyle Zollo-Venecek

Kyle.Zollo-Venecek@tuftsmedicine.org

tuftsmedicine.org