Stamen Health

Stamen Health — Strategic Positioning and Market Opportunity. EU private hospital EHDS compliance and Personal Health Knowledge Graph (PHKG) infrastructure from Oslo, Norway.

For a one-page executive summary, see Stamen Health Executive Summary.

Vision

Stamen Health builds the EHDS compliance layer and PHKG infrastructure for EU private hospitals — starting from Norway, expanding across the EU. We turn fragmented, heterogeneous hospital data into structured, ontology-backed knowledge graphs that serve patients, clinicians, and researchers simultaneously.

From Oslo to Europe: curate once, reuse many — at commercial scale.

Starting Point: AIDAVA

AIDAVA (EU Horizon Europe, Grant 101057062, EUR 7.7M, Sep 2022 — Aug 2026) is the only research project that has built a full end-to-end pipeline for Personal Health Knowledge Graphs:

Heterogeneous data ingestion (structured + unstructured)
NLP extraction from clinical narrative in multiple languages (Dutch, German, Estonian)
PHKG creation using SNOMED CT, HL7 FHIR, LOINC ontologies
Automated FAIRification
Patient-facing explainable AI
Multi-stakeholder reuse (patients + clinicians + researchers)

AIDAVA's honest result (March 2025 evaluation): 45% of documents curated automatically. 20 minutes per document. Usability good, but explanations suboptimal. G2 delivery end 2025, testing early 2026. Project ends August 2026 with a research prototype, not a commercial product.

Stamen Health's thesis: AIDAVA's research architecture is correct. The gap is commercialization speed and production-grade engineering. A well-funded Norwegian startup with AIDAVA's team connections and the right co-founders can take this architecture, harden it, and sell it to EU private hospitals — starting NOW, ahead of the EHDS compliance wave.

The Market Opportunity

EHDS Compliance Wave (2026-2030)

The European Health Data Space (EHDS) regulation mandates that every EU hospital make health data available in standardized, interoperable formats by 2029-2030.

The compliance timeline:

2025-2026: National transposition into EU member state law
2027-2029: Hospital infrastructure build-out
2029-2030: Mandatory data availability

Every EU hospital needs EHDS compliance tools. Every private hospital chain needs them faster (competitive pressure). This is a multi-billion euro market — similar to GDPR compliance in 2018-2021, but for health data.

The PHKG Infrastructure Bet

Personal Health Knowledge Graphs are the right architecture for longitudinal health data. Unlike relational databases or flat FHIR bundles, PHKGs:

Represent complex clinical relationships over time
Support ontology-based reasoning (SNOMED CT hierarchy)
Enable cross-system queries that flat data cannot
Scale for AI/ML downstream (clinical NLP, decision support, trial matching)
Serve multiple stakeholders from the same graph ("curate once, reuse many")

The market for knowledge graphs is $6.9B by 2030. Healthcare is the fastest-growing vertical. No current player has a PHKG-specific product for EU hospitals.

Competitive Landscape

What Exists Today

Competitor	Country	What They Do	Critical Gap for Stamen
PicnicHealth	US	Patient-anchored medical records, 10K+ facilities, $60M+ raised	US-only, no knowledge graphs, no FAIRification
Better	Slovenia	Open-source FHIR platform	No AI curation, no NLP, infrastructure not intelligence
InterSystems	US/EU	HealthShare in 100+ countries, ~$1B+ revenue	Enterprise infrastructure, no automation, no patient-facing explainability
Castor EDC	Netherlands	Clinical trial FAIRification, 10K+ studies	Trials only, not hospital data, no NLP
Cogstack	UK	NHS clinical NLP (open-source)	NHS-specific, no knowledge graphs, no FAIRification
Healx	UK	Knowledge graphs for drug discovery, $47M+ raised	Drug repurposing, not patient records, no hospital data
1upHealth	US	FHIR patient platform, $40M raised	Data access layer, no curation, no NLP, no KG
Averbis	Germany	German clinical NLP	One language, no KG, no hospital integration
Qantev	France	AI claims processing, €30M raised (2025)	Insurance claims, not hospital records, no KG
Owkin	France	Federated learning, $300M+ raised	Model training, not data curation, no KG
LynxCare	Belgium	Clinical data platform, real-world evidence	No KG, no NLP, hospital-focused but not PHKG
AIDAVA (research)	EU	Full PHKG pipeline prototype	Research only, ends Aug 2026, no commercial product

The Gap Stamen Fills

No current competitor offers EHDS-compliant PHKG infrastructure purpose-built for private hospitals, production-grade automated curation (AIDAVA reached 45%, target 80%+), multi-language NLP (Norwegian, Swedish, Danish, then German/English), SNOMED CT ontology-backed knowledge graphs, patient-facing explainable AI for health record understanding, and "curate once, reuse many" for private hospital chains.

Who Could Close the Gap

PicnicHealth* — could add FAIRification and KG on top of US data, but US-only and no EHDS angle
Better* — could add AI curation layer, but Slovenian/enterprise sales motion is slow
Owkin* — could add patient-facing features with $300M, but federated learning is a different architecture bet
InterSystems* — could add automation, but enterprise sales cycles are 12-18 months, no startup speed
Google Cloud / Microsoft* — could dominate with FHIR APIs, but hospitals distrust big tech and EU regulatory complexity

Stamen's advantage: Startup speed + AIDAVA research foundation + Norwegian EHDS leadership + EU private hospital focus.

Stamen Health's Position

First Move: EHDS Compliance Infrastructure

Target customers: Private hospital chains in Norway, Sweden, Denmark, then Germany/Netherlands.

Value proposition: "We make your hospital EHDS-compliant in 12 months, not 36 months. Your data becomes structured, interoperable, and AI-ready from day one."

Products: EHDS Readiness Assessment (audit current data maturity against EHDS requirements), PHKG Pipeline (automated curation of heterogeneous hospital data into SNOMED CT-backed knowledge graphs), Compliance Dashboard (ongoing monitoring against EHDS mandates), and Data Export API (FHIR-native data availability for EHDS MyHealth@EU cross-border access).

Pricing: SaaS subscription (per bed / per hospital) + implementation fees. EUR 50K-200K for implementation, EUR 10K-50K/year for subscription.

Second Move: Clinical Intelligence Layer

Once PHKG infrastructure is deployed, add Clinical Decision Support (doctor sees complete longitudinal patient history with SNOMED CT-coded problem list), NLP-powered Discharge Summary (automated generation from structured + unstructured data), Trial Matching (patient-to-clinical-trial eligibility matching using PHKG), and Research Data Service (de-identified, FAIRified datasets for pharma/academic research).

Third Move: Patient-Facing PHR

Private hospital-branded patient app built on PHKG: Complete longitudinal health record (from all hospital encounters), explanation of diagnoses and medications in plain language, consent-based data sharing for second opinions or research, and preventive health nudges based on longitudinal patterns.

Curation Accuracy — The Central Technical Question

AIDAVA achieved 45% with 2022-era tools. Modern LLMs plausibly improve this materially — perhaps 60-85% depending on document type and language. Whether 90%+ is achievable is an open empirical question.

The economics are highly sensitive to this number. At 70%+ per-fact recall with HITL, the pitch is "makes curation 3x faster with 0.5 FTE oversight." Below 60%, the thesis doesn't work with current technology.

For the full analysis — accuracy definitions, the compounding problem, HITL costs, what the binding constraints actually are for each opportunity, and the year-1 feasibility study plan — see Stamen Health Executive Summary#Curation Accuracy — The Central Technical Question.

Why Norway, Why Oslo

EHDS implementation leader: Norway is among the first EU/EEA countries implementing EHDS, with strong national health data infrastructure (KRR, e-Helse)
Digital health talent: Norway has 15+ years of health IT development, e-health startups, and FHIR adoption
Clinical NLP expertise: AIDAVA connections + access to Norwegian clinical text for NLP training
Trust advantage: Norwegian hospitals trust Norwegian vendors over US big tech — and EU hospitals trust Norwegian companies (GDPR-conscious, not NSA-adjacent)
Soft funding landscape: Innovation Norway grants, SkatteFUNN, IPN — non-dilutive capital available for EHDS-related R&D
Nordic expansion path: Norway → Sweden → Denmark → Finland, then DACH and Benelux

Norwegian Health Data — The Strategic Window (April 2026)

At Helsedatadagen (Ehin, April 10, 2026), Health Minister Jan Christian Vestre issued a direct challenge to the Norwegian health tech sector:

"Det nytter ikke bare å konstatere at vi har spiren til noe vakkert her, vi må faktisk bruke det til noe." (It's not enough to note that we have something beautiful here — we must actually use it.)

The Warning

Vestre called Norwegian health data "the gold" but warned the advantage is being squandered:

Norway talks more about the data than it uses it. The gap between rhetoric and action is growing.
Processing time is too slow: Helsedataservice applications increased ~30% in 2025, but capacity hasn't kept pace. Manual procedures bottleneck data release.
Other countries will exploit the data if Norway doesn't: "If we don't see the gold, others will."
Industry must take more risk: "This is a march order to industry — take more risk, be willing to use more capital and invest more in Norway."
Clinical trial demand is rising: 143 applications to DMP in 2025 (up from ~120 in prior years). Action plan for clinical studies published.

Source: MedWatch, Vestre at Helsedatadagen (Apr 10, 2026). Full article: Norwegian Health Data Infrastructure.

Why This Matters for Stamen Health

This is the exact market timing Stamen needs.

The minister's warning creates a regulatory pull:

Government wants data to flow faster — means hospitals need infrastructure to release structured data. Stamen's EHDS compliance layer is that infrastructure.
AI for data automation is now official policy — "use KI and de-bureaucratize the process." Stamen's ontology-backed data structuring is exactly this.
Clinical trial demand rising — more pharma companies want Norwegian data. Stamen can be the middleware that makes hospital data trial-ready.
Nordic Plus cooperation coming — cross-border health data sharing for rare diseases. Stamen's standardized PHKG architecture is designed for interoperability across borders.
Industry told to invest — Vestre explicitly called for more risk capital in health data. Government signals > regulatory certainty.

The Timing Argument

"Bedrifter skal drives for egen regning og risiko. Den viktigste rammebetingelsen vi har er å ha tilgang til de beste helsedataene i verden." (Vestre)

The minister's logic: Norway has the data → government is opening access faster → industry must build on it now. If Stamen starts building the EHDS compliance layer in 2026, it's aligned with:

EHDS implementation timeline (2026-2030)
Helsedataservice capacity expansion
Nordic Plus data sharing agreements
EU Digital Europe EHDS funding calls

Waiting means competing with larger players who will enter as data access improves.

EU Expansion Strategy

Phase 1: Nordic (2026-2027) — 2-4 private hospital groups in Norway as anchor customers, 1-2 Swedish or Danish private hospital pilots, build Norwegian clinical NLP models.

Phase 2: DACH + Benelux (2027-2028) — German private hospital chains (medium-sized, not Charite-sized), Dutch private hospitals and clinics, multi-language PHKG (Norwegian + German + Dutch).

Phase 3: EU-wide (2028-2030) — EU expansion through partner channels, PHKG infrastructure as platform for pharma research data, patient-facing PHR at scale.

Revenue Model

EHDS Compliance SaaS — subscription per hospital
Implementation Services — one-time setup + customization
Clinical Intelligence — premium layer on top of PHKG
Research Data Access — pharma/academic licensing of de-identified PHKG data

Year 1-2: Implementation + SaaS (B2B). Year 3-4: SaaS + Clinical Intelligence. Year 5+: Platform (data services + PHR).

Competitive Moat

AIDAVA-derived architecture: PHKG ontology design, NLP pipeline, FAIRification approach — validated by EUR 7.7M research grant
EHDS compliance complexity: The regulation is 100+ pages of technical requirements — building expertise is a 2-3 year head start
Clinical NLP in Norwegian/Swedish: Low-resource language clinical NLP is not trivial; first mover advantage
SNOMED CT expertise: Ontological reasoning over longitudinal data requires deep SNOMED CT knowledge
Hospital trust: Private hospitals want a partner, not a vendor — relationship-based selling favors regional players
Data network effects: Each hospital PHKG improves the ontology model and NLP for all customers

Risks

AIDAVA IP if consortium IP claims are unclear — need IP agreement early
InterSystems / big tech moves fast — but enterprise sales cycles are long, and hospitals want alternatives
EHDS timeline slips — but the mandate is already law, delays compress rather than eliminate demand
Finding the right co-founders — need COO with hospital relationships and CCO for commercial expansion
Regulatory complexity — MDR, IVDR, GDPR叠加 EHDS — need strong regulatory affairs from day one

Team Requirements

What Stamen needs to build:

CTO / Technical Co-founder: Deep expertise in clinical NLP, knowledge graphs, FHIR — ideally from AIDAVA or similar project
COO / Norwegian Co-founder: Hospital relationships, operational delivery, Norwegian health system knowledge
CCO (Year 2+): Commercial leader with EU hospital sales experience
Clinical NLP Engineer: Norwegian/Swedish clinical text models
Knowledge Graph Engineer: SNOMED CT, FHIR, ontological reasoning