CDE Harmonization

This project makes Common Data Elements (CDEs) more interoperable across clinical research studies using LinkML schemas and AI-assisted semantic mapping.

What We're Doing

Clinical research uses CDEs—standardized data fields with defined permissible values—but they're fragmented across repositories and lack semantic bindings to ontologies. This limits data integration and AI-readiness.

We're building tools to:

• Collect CDEs from major repositories (NIH, PhenX, caDSR, RADx, HEAL)
• Convert to LinkML schemas for computability
• Generate semantic mappings using AI and human curation
• Enable data harmonization across studies

Documentation

📖 Full documentation: https://monarch-initiative.github.io/cde-harmonization/

• Installation Guide
• Schema Documentation
• Project Roadmap

Repository Contents

• data/ - Raw CDEs from multiple repositories
• linkml/ - Generated LinkML schemas
• cde2linkml/ - Conversion tools
• docs/ - Documentation source

Technologies

• LinkML - Schema modeling framework
• SSSOM - Mapping standard
• Ontologies - LOINC, HPO, Mondo, NCIT, OBA