MCP server for mapping clinical terminology to Observational Medical Outcomes Partnership (OMOP) concepts using large language models for standardizing medical data.

Installation

From Source Code

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

Configuration

Claude Desktop

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

Available Tools

Tool	Description
find_omop_concept	Maps clinical terminology to OMOP concepts, validates terminological mappings, searches the OMOP vocabulary...

Features

• Mapping clinical terminology to OMOP concepts
• Validation of terminological mappings
• Searching the OMOP vocabulary
• Conversion between different clinical coding systems

Environment Variables

Optional

• AZURE_OPENAI_ENDPOINT - Azure OpenAI service endpoint
• AZURE_OPENAI_API_KEY - Azure OpenAI API key
• AZURE_API_VERSION - Azure API version
• MODEL_NAME - Azure OpenAI model name
• OPENAI_API_KEY - OpenAI API key (alternative to Azure)

Usage Examples

Map `Temperature Temporal Scanner - RR` for `measurement_concept_id` in the `measurement` table.

Map clinical terms with preferred vocabularies in order of priority (e.g., "SNOMED preferred" or "LOINC > SNOMED > RxNorm")

Notes

Requires the uv package manager. It is recommended to specify the field and OMOP table name in prompts to improve accuracy. Refer to omop_concept_id_fields.json for a list of OMOP fields and tables that store concept identifiers. Environment variables are only needed for API calls.