.jpg)
One Meaning for Clinical Data, Across Every Study
.jpg)
The Challenge
A global pharma organization ran multi-study, multi-vendor clinical programmes. The data existed. Getting it to mean the same thing in every study, system and analysis was the problem.
Key Challenges Included:
- Same Data Re-Mapped at Every Study Start. Vendor, laboratory and CRO feeds were re-mapped study by study. Mapping knowledge sat with individual programmers, and integration issues surfaced late, at data review.
- Cohort and Eligibility Logic Rebuilt Each Time. Disease, phenotype and lab criteria were written differently in protocols and data, so feasibility depended on manual pulls and spreadsheets.
- No Traceability for Standards and Terminology Versions. Controlled terminology lived in documents apart from the data. Teams could not show which version a study used, or who approved a definition.
- Pooled Analysis and GenAI Without Shared Definitions. Legacy studies, safety terms and AI pilots each harmonized the same concepts again, with no guarantee the answers were consistent.
Our Solution
A Governed Clinical Ontology, Defined Once and Reused Everywhere
Standards-Anchored Clinical Vocabulary
Eligibility, endpoint, lab, unit and condition concepts anchored to CDISC Controlled Terminology, NCIt, MONDO, HPO, LOINC and UCUM (MedDRA and SNOMED CT where licensed).
AI-Assisted Alignment with Human Approval
EDC, lab, ePRO and CRO sources profiled and proposed for alignment; data managers and standards librarians review and approve every mapping.
Reusable Semantic Models and Mappings
Approved cohort entities, endpoint definitions and source mappings published once, so each new study builds on earlier work instead of starting blank.
Governed Lifecycle and Inspection Readiness
Review workflow, versioned releases, lineage, audit and role-based access record who approved each definition and what changed between releases.
Grounded Clinical AI
Business-language questions answered from governed definitions with provenance, and ontology context provided to AI agents through APIs and MCP.
The Outcomes
One Definition, Every Study
Eligibility, endpoint and lab concepts defined once and reused, so feasibility and cohort questions start from shared definitions.
Study Start-Up That Builds on Earlier Work
Vendor and CRO mappings become reusable assets, and integration issues surface at onboarding rather than at data review.
Traceability Inspectors and QA Can Follow
Versioned, approved definitions with lineage and audit replace scattered documents and email threads.
Cross-Study Reuse and Trusted Clinical AI
Pooled analyses and GenAI answers start from a shared, governed semantic model.

.jpg)
.jpg)
.jpg)
.jpg)