Edence Resolve: AI-Powered OMOP Harmonisation for EHDS
Discover Edence Resolve, an AI-assisted tool for OMOP data harmonisation. Scale your healthcare data engineering and prepare for EHDS compliance.
Edence Resolve
AI-assisted data harmonisation for OMOP
An edenceHealth product
Pitch for Sam Bambust & Tine Verlonje
May 2026
THE CONTEXT
The OMOP wave is coming — and the workforce isn't ready
The European Health Data Space goes live by 2029.
Every health data holder will need to harmonise to OMOP. Manual ETL won't scale — there simply aren't enough data engineers.
"The question is no longer if — it's how fast."
02
WHY NOW
The EHDS shifts the math
EHDS + Darwin-EU drive a step-change in OMOP harmonisation demand
Pharma under cost pressure — every euro per pipeline scrutinised
Vendors, registries and hospitals must industrialise OMOP delivery
The current bottleneck is human, not technical
03
FOR WHOM WE BUILD
Built for the people carrying the weight
Pharma R&D, Hospitals & Registries
Must deliver OMOP-conformant data at scale
Data Engineering Teams
Run the pipelines, carry the technical burden
Business Owners
Carrying the cost and lead time pressure
edenceHealth Analysts
Our own team — proof point, using it on live client work
THE PRODUCT
One workspace.<br/>Three AI modules.
End-to-end OMOP delivery — from source understanding to production SQL
Profiler
Understand the source
Mapper
Match to OMOP vocabularies
Generator
Produce the SQL
Pre-flight Check
Validator
Release
05
MODULE 01
PRODUCTION READY
Profiler
Production-grade source intelligence — running on live client work today
What it does:
Reads source database
→ structured mapping document
Flags logical relationships,
quality issues, ambiguities
Runs on local model
— patient data never leaves environment
Built-in safeguards
keep sensitive data out of outputs
Human validation
as final safety net
● LIVE IN PRODUCTION
MODULE 02
Mapper
Semantic matching from source terms to OMOP concepts — with confidence scoring
STABLE & PRODUCTION READY
What it does:
Maps source terms to OMOP concepts with confidence scoring
Combines Lucene fuzzy matching with transformer reasoning
External benchmark: solid mid-pack on strict scoring
Stronger in real-world use — multiple suggestions = more value
Benchmark result
In external mapping competition — mid-pack on strict single-answer. In real-world use, where multiple plausible suggestions matter, the advantage is clear.
07
08
MODULE 03
BETA
Generator
Two deployment options, both transparent. The choice belongs to the client.
Local Model
Data stays on-site
✓ Maximum privacy
⚡ Lower SQL quality
Cloud — AWS Bedrock
Medically certified infrastructure
✓ Higher SQL quality
⚡ Requires data handling agreement
No public LLMs. No black boxes. No surprises.
ROADMAP
We build the cupcake first — then the wedding cake
Today
Cupcake release — Summer 2026
Log in to Edence Resolve
Three working modules
User validates manually
Workflow ~30% faster already
Full Vision
The wedding cake
Automated pipelines
Accessible beyond data engineers
Streamlined collaboration (dev + clinical)
Scale without headcount growth
THE PAIN TODAY
Before EHDS volume even hits
€30–40K
per OMOP ETL project
10×/year
typical annual pipeline volume
20 days
analyst days per pipeline
That's ~€350,000 per year — before EHDS-driven volume multiplies demand
And this is before EHDS-driven volume hits
THE IMPACT
Speed is the differentiator. Quality is table-stakes.
20 days → 10 days
Lead time per ETL — realistic, not aspirational
~€100K/year freed
At €25k per project, reinvestable at constant volume
Unlimited throughput
At EHDS scale — manual delivery simply cannot match
The role becomes more strategic — not redundant
THE ANALYST EVOLUTION
AI does the mechanical work. Analysts become architects.
Mechanical
Before Resolve
Writer of SQL, mapper of terms, manual profiler
Strategic
With Resolve
Reviewer, validator, quality gate
Leadership
At Scale
Architect of data pipelines, clinical collaborator
Audit-logged
Explainable AI
Scales without headcount
Every step traceable
Throughput scales. Headcount doesn't have to.
TRUST & COMPLIANCE
Built for regulated environments. By design.
Local Deployment
Sensitive data stays on-site. No external transmission.
AI Act Ready
Explainability and traceability built in by default — not retrofitted.
Certified Cloud Option
AWS Bedrock — medically certified. No public LLMs, ever.
Pull-by-Client Release
We publish a release tag. Client repo hook pulls it in. We never touch production.
"We never touch the client's production environment."
COMPETITIVE EDGE
The market is service-dominated. We're the product.
No direct productised competitor exists today. That window won't stay open indefinitely.
01
End-to-end pipeline automation
Not a point solution — the full journey
02
Clinical-grade validation built in
Not bolted on — designed from day one
03
B2B product, not services-with-AI
Scalable economics. Reproducible delivery.
04
No productised competitor
Service-dominated market — we're first
05
Proven on live client work
edenceHealth analysts use it today
The window won't stay open indefinitely →
Traction
Where we stand today
2 Co-development Partnerships
Funding and shaping the product in real use
Live on Client Engagements
edenceHealth analysts use Resolve on actual projects today
J&J Informal Interest
3 pipelines via the platform — warm signal from a global pharma leader
Profiler in Production
Engine live and proven. SQL Generator in beta with iteration.
Cupcake Release
Targeted for late summer 2026
THE ASK
Not 'buy this.'
Help us prove it.
3–5
Conversations to open in your network — J&J accounts, Cronos health data clients, EHDS-aligned initiatives
You open the door
We bring the demo, the materials and the technical depth
July–August 2026
Validation conversations to get started immediately
We do the heavy lifting. You make the introduction.
16
NEXT STEPS
Three milestones. One clear path.
July – August 2026
Validation Conversations
3–5 warm introductions. We demo, you open the door.
End of Summer 2026
Cupcake Beta Release
Full working product. Three modules. Real clients.
Q4 2026
Go / No-Go Decision
Broader scale-up decision based on validation evidence.
First step today:
agree which 3–5 conversations make sense to set up.
Thank you.
edenceHealth — Edence Resolve
Serge Hufkens
Isaac Claessen
Freija
Panos
Lars
edencehealth.com
Questions? Let's talk.
- omop
- health-data
- ai-healthcare
- ehds
- data-harmonisation
- medtech
- health-informatics