Back to Changelog
Feature

OMOPHub MCP Server: Medical Vocabularies for AI Assistants

New MCP Server gives AI assistants like Claude direct access to 10M+ OHDSI ATHENA medical vocabulary concepts. No database setup or CSV downloads required.

Today we're releasing the OMOPHub MCP Server (@omophub/omophub-mcp): a Model Context Protocol server that gives AI assistants instant access to OHDSI ATHENA medical vocabularies. Search concepts, map between coding systems, and explore hierarchies directly from your AI workflow, without setting up a database or downloading CSV files.

Available Tools

The MCP server exposes six tools that cover the most common vocabulary operations:

ToolDescription
search_conceptsSearch for medical concepts by name, code, or keyword across all vocabularies
get_conceptLook up a specific concept by its OMOP concept ID
get_concept_by_codeFind a concept using its source vocabulary code (e.g., ICD-10 code E11.9)
map_conceptMap a concept from one vocabulary to another (e.g., ICD-10 to SNOMED)
get_hierarchyExplore parent/child relationships and ancestor/descendant trees
list_vocabulariesList all available vocabularies with concept counts and metadata

Supported Clients

The MCP server works with any MCP-compatible client:

  • Claude Desktop - native MCP support via config file
  • Claude Code - add with claude mcp add
  • Cursor - MCP configuration in settings
  • VS Code - via Copilot MCP extension
  • Docker - containerized deployment for teams
  • Streamable HTTP - connect from any HTTP-capable client

Quick Start

Run the server directly with npx - no install needed:

npx -y @omophub/omophub-mcp --api-key YOUR_API_KEY

To add it to Claude Desktop, add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "omophub": {
      "command": "npx",
      "args": ["-y", "@omophub/omophub-mcp"],
      "env": {
        "OMOPHUB_API_KEY": "YOUR_API_KEY"
      }
    }
  }
}

For Claude Code, add the server with a single command:

claude mcp add omophub -- npx -y @omophub/omophub-mcp --api-key YOUR_API_KEY

Example: ICD-10 to SNOMED Mapping

Once connected, your AI assistant can work with medical vocabularies conversationally:

You: What SNOMED concept does ICD-10 code E11.9 map to?

Assistant: ICD-10 code E11.9 ("Type 2 diabetes mellitus, without complications") maps to SNOMED concept 201826 - "Type 2 diabetes mellitus". This is a Standard concept in the Condition domain, making it suitable for use in OMOP CDM analyses.

No manual lookups, no CSV files, no database queries - just ask.

Resources