OMOP MCP Server maps clinical terminology to Observational Medical Outcomes Partnership (OMOP) concepts using large language models, helping standardize medical data across different clinical coding systems (like SNOMED, LOINC, and RxNorm).

Use it when you need to convert a clinical term (a lab test name, measurement label, or diagnosis description) into its corresponding standardized OMOP concept for a specific OMOP table and field - for example, mapping "Temperature Temporal Scanner - RR" to a measurement_concept_id. Specifying the target field and OMOP table name in your prompt improves mapping accuracy; refer to omop_concept_id_fields.json in the repo for the list of OMOP fields and tables that store concept identifiers. You can also specify a preferred vocabulary priority order (e.g., "SNOMED preferred" or "LOINC > SNOMED > RxNorm"). Do not use it without an LLM backend configured - it requires either Azure OpenAI or OpenAI API credentials to perform the mapping.

• find_omop_concept: maps clinical terminology to OMOP concepts, validates terminological mappings, searches the OMOP vocabulary, and converts between different clinical coding systems.

git clone https://github.com/OHNLP/omop_mcp.git
cd omop_mcp

Requires the uv package manager. Configure Claude Desktop:

{
  "mcpServers": {
    "omop_mcp": {
      "command": "uv",
      "args": ["--directory", "<path-to-local-repo>", "run", "omop_mcp"]
    }
  }
}

Set API credentials via environment variables, either Azure OpenAI (AZURE_OPENAI_ENDPOINT, AZURE_OPENAI_API_KEY, AZURE_API_VERSION, MODEL_NAME) or OpenAI (OPENAI_API_KEY) as an alternative.

Healthcare informaticists and researchers working with OMOP Common Data Model datasets who need to standardize clinical terminology into OMOP concepts for observational research or data harmonization.