COHA ONE HEALTH DATASETS INITIATIVE

Progress and Plans

​

Joe Strecker

ACTIVITIES

• Workshops
• Governance Team, Technical Proof-of-Concept Team
• Data discovery, software development
• Developing conventions and best-practices for implementations
• Connecting across veterinary informatics communities
• Keeping an eye out for collaboration and funding opportunities

​

​

​

​

2018 FEB WORKSHOP – 2 days�Organized by Sue VandeWoude, Elle Holbrook

Aims: gauge interest, brainstorm, generate momentum

​

Attendees:

• Nearly 40
• 27 academics representing 13 institutions
• Morris Animal Foundation
• VetCOT and VMDB data collections
• Mars PetCare and member organizations (e.g., Banfield, Blue pearl)

​

Talks / Panels:

• EHR data model use in human med, adaptations to veterinary med
• Other veterinary data networks (e.g., VMDB, VetCOT)
• Sustainability of networks, use cases, next steps

​

​

​

​

​

​

2018 FEB WORKSHOP

Main Outcomes:

​

• Increased awareness and interest of challenges and opportunities
• Need for technical and governance committees to steer further efforts
• Review state of veterinary informatics - published
• Awarded COHA Pilot grant to continue to organize and build
• Launchpad

​

​

​

2019 FEB GOVERNANCE/TECHNOLOGY WORKSHOP – 3 days�

Aims: Foundational mtg, deeper dives into Tech and Gov

​

Attendees: 18, mostly technical but also governance core

​

Topics:

• Governance: Facilitated by Virginia Rentko, Kelly Hall and Sarah Moore
• Purpose, guiding principles
• Organizational structure
• Handling data requests
• Publication guidelines
• Technical
• Deeper dive into OMOP
• Academic and Industry EHR/Data Implementations
• Data mapping and loading processes and tools
• Measuring data quality
• Learning from other data sharing implementations

​

​

​

​

​

​

​

• Draft of governance document for TRANSLATOR
• asd

​

​

​

​

2019 FEB GOVERNANCE/TECHNICAL WORKSHOP – 3 days

Main Outcomes

​

• Draft Governance document
• Decision to move forward with OMOP as CDM
• Establishment of a Technical Proof of Concept team:
• Tufts University – Manlik Kwong
• UC Davis – Chris Brandt, Allison Zwingenberger
• CSU – Joe Strecker

​

​

​

​

​

​

​

​

TRANSLATOR Steering Committee (SC)

• One member from each site (site representative; SR)
• Meets Quarterly

​

​

​

Governance Subcommittee

Science and Data Use Core

Data Infrastructure Core (DIC)

• SRs assure
• Consistent participation of sites
• Local regulatory approvals in place
• Raise awareness of TRANSLATOR resource
• Address scientific misconduct

Management Core

• Assist sites with local regulatory processes
• Secures data use agreements
• Executes and oversees contracts

• Researchers from TRANSLATOR sites
• Advises on cohort development and study designs
• Reviews proposed study protocols
• Provides recommendation for approval by SC
• Works with DIC on development of new content areas

• Advises SC on development, implementation and expansion of TRANSLATOR network
• Includes investigators and technical data experts
• Implements OHSDI and OMOP CDM
• Supports site implementation via technical guidance
• Ensures technical best practices
• Standardizes data query creation and data QA/QC processes

• Develop and implement policies and procedures for TRANSLATOR decision-making
• Advise and make recommendations to the SCI and other cores as appropriate

2020 OCT GOV/TECH WORKSHOP – 2 day

Aims: Outreach to broader COHA, lessons from POC technical efforts

​

Attendees:

• 55 registrants, 18 speakers
• Mix of academics including several COHA veterinary institutions, clinical trials, IT
• Morris Animal Foundation, 1-Data, VetCOT, AVI
• Data services, veterinary interests (Odysseus, VCA, Banfield, etc)

​

Topics:

• Reinforcing previous technical sessions, introducing to more academics
• OMOP use in the real world
• Challenges / successes of translating vet data to OMOP
• Veterinary analytics in non-academic settings
• Collaboration opportunities
• Governance discussions/breakouts

​

​

�POC STATUS - CSU

Legacy and production EHR systems

• Over 100K visits manually encoded before 8/2018 – proven potential for machine learning, 3 publications
• New, commercial StringSoft EHR less transparent, limited encoding and querying capability, open database

​

Common ETL framework developed

• Built with reusability in mind but significant adaptation required
• Legacy data model is well-understood and validated
• ETL has been run against legacy, works but needs refinement

​

Production data model is mostly understood

• ETL is under development, framework still valid

​

Technical workforce is students

• Bright, talented, engaged, inexpensive, and temporary...
• ETL code, infrastructure, documentation getting heavier with progress
• Harder to develop institutional knowledge

​

​

​

​

​

GROWING THE NETWORK

Interest is strong

• Value to academia – actionable operational data, local research, network research
• Integration with human healthcare
• Value to industry

​

OMOP is a heavy upfront lift to encode, but can be worth it

• Drugs, procedures, conditions – use mappings to auto-encode
• Automate extraction of meaning from unstructured text – machine learning
• Well structured interfaces in EHR to facilitate fast, detailed data entry by clinicians
• Start with most tractable data – it has value
• Simple sharing model, point-to-point DUAs
• Establish baseline of data harmonization and quality

​

​

​

​

​

​

OPPORTUNITIES AND NEXT STEPS

POC team willing to share our journey with others

• Goals align with many institutional needs, aspirations
• Grow our experience to make increasingly successful funding proposals
• We are enjoying this!

​

Develop and maintain connections

• This group
• COHA
• Association for Veterinary Informatics (AVI)
• Veterinary Terminology Services Lab (VTSL)
• OHDSI/OMOP
• Veterinary Machine Learning Group

​

​

​

THANKS!

• Monarch Initiative, Translational Integrative Sciences Lab (TISLab)
• Colorado Clinical and Translational Sciences Institute (CCTSI)
• Sue VandeWoude – CSU One Health Institute
• CTSA One Health Alliance (COHA)
• Terri Ward – Veterinary Teaching Hospital (VTH) Medical Records
• Jamie Joy, Ryan Nowell – VTH StringSoft administrators
• Sally Winden, Paul Norvelle, Hussein Al Rousan - CVMBS IT Services
• Dawson Eliasen, Sooraj Lankala, Adam Kiehl, Maria Samson, James Rudd, Erik Gary

​

​

​

GOVERNANCE CONSIDERATIONS

Access to data

• Open data
• COHA-only, approved request-only
• Point-to-point

​

Location of data

• Centralized vs distributed

​

Anonymization, patient/owner privacy

• Aggregated vs. individual case data
• Impact of certain infectious disease data

​

Appropriate use

• Attribution for publication
• Commercialization, grants

​

​

​

​