Skip to content

Custom Concepts

Custom concepts are additions to the standard OMOP vocabulary that capture Emory-specific data elements not covered by existing OHDSI ontologies. These allow Emory researchers to query local data elements using the same OMOP patterns used for standard concepts.

How Custom Concepts Work

The Enterprise OMOP team maps source values from Epic and CDW to standard OMOP vocabulary concepts wherever possible. When no standard concept exists for a clinically important data element, a custom concept is created in Emory's local vocabulary extension.

Custom concepts:

  • Follow OMOP naming conventions and are assigned concept IDs in a reserved range (2 billion+)
  • Are queryable using the same concept, concept_ancestor, and concept_relationship tables as standard concepts
  • Are documented and versioned alongside each product release
  • Are built through the Custom Vocabulary Builder (CVB), an automated pipeline that produces OMOP-compatible vocabulary deltas

Architecture

  • Custom Vocabulary Strategy

    Why Emory adopts 2-billionaire Standard concepts, how STCM and CVB complement each other, and our early adoption of vocabulary metadata tables.

    Vocabulary Strategy

  • Network Study Bifurcation

    How the ETL supports both local (CVB-enhanced) and OHDSI network study (Athena-only) concept resolution through compound targets and dual vocabulary schemas.

    Network Study Bifurcation

How to Get Involved

  • Request a Mapping

    Found unmapped source data in Epic or CDW? Submit a request and the vocabulary team will triage, prioritize, and build it into the next release.

    Requesting Mappings

  • Contribute a Vocabulary

    Have domain expertise and want to contribute mappings directly? Use the CVB pipeline to build OMOP-compatible vocabulary deltas and submit via pull request.

    Contributing Vocabularies

Current Coverage

The Enterprise OMOP team has prioritized mapping the most common source values across ingested domains. However, coverage is not exhaustive — particularly for:

  • Uncommon clinical terms not tied to a standard medical ontology in Epic
  • Data elements from specialized workflows or research-specific instruments
  • Non-standard identifiers and local coding systems

If you believe important data is missing, request a mapping to start the process.