PHKG Business Opportunities

Detailed analysis of five commercial opportunities for Personal Health Knowledge Graph (PHKG) infrastructure. Each builds on the same core technology — curating fragmented health data into structured, ontology-backed knowledge graphs — but targets different customers and revenue streams.

See also: Stamen Health, Stamen Health Executive Summary, PHKG Business Models & Market, AIDAVA.

Overview

The PHKG architecture from AIDAVA enables five distinct business opportunities, each with different customers, timelines, and risk profiles:

Opportunity	Customer	Revenue Model	Time to Revenue	Difficulty
1. Data Intermediary	Pharma/CRO	Commission on data access	18-24 months	High (regulatory)
2. B2B Preventive Health	Employers	Per-employee subscription	12-18 months	Medium (sales)
3. Patient Group Apps	Patients (rare/chronic)	Subscription SaaS	6-12 months	Medium (adoption)
4. GP Diagnostic Overview	GP practices	Per-practice subscription	12-18 months	High (integration)
5. Pre-Consultation Triage	GP practices	Per-consultation fee	6-12 months	Medium (simplest product)

The core PHKG technology serves all five, but each requires different product packaging, go-to-market strategy, and regulatory approach.

1. Health Data Intermediary — Hospitals Sell to Clinical Trials

Concept: Hospital collects and curates patient data into PHKG. Pharma companies pay for access to curated, de-identified patient cohorts for clinical trial recruitment, real-world evidence generation, or post-market surveillance.

How It Works

1. Hospital data (structured + unstructured) is curated into PHKG using Stamen's platform
2. Patients consent to data sharing for research (GDPR-compliant, EHDS-aligned)
3. Pharma/CRO searches the PHKG for cohorts matching trial inclusion criteria
4. Hospital receives payment per patient enrolled or per dataset accessed
5. Stamen takes a commission (platform fee) on each transaction

Market Context

Existing players:

• Datavant (US, $7B valuation 2021) — data linking and de-identification for US hospitals. Dominant in US, no EU presence.
• TriNetX (US/EU, $500M+ valuation) — federated network of 300+ organizations. Grows data, connects to pharma. Focuses on trial feasibility.
• LynxCare (Belgium, ~€50-80M) — generates real-world evidence from hospital data. Connected to Belgian hospitals.
• IQVIA (US, $35B market cap) — clinical data services. The incumbent everyone competes with.

What's different about PHKG approach:

• TriNetX and IQVIA work with structured data (ICD codes, lab values). They miss unstructured clinical notes.
• AIDAVA's NLP extracts value from discharge notes, radiology reports, specialist letters — the 80% of clinical data that's currently unstructured.
• PHKG with SNOMED CT coding enables more precise cohort matching than flat ICD-code searches.

Revenue Potential

• Per-patient-enrolled fee: €500-2,000 per patient enrolled in a clinical trial through the platform
• Dataset access fee: €10,000-50,000 per curated dataset for feasibility studies
• Commission rate: 15-30% of hospital payment

Realistic year 2 revenue: €100-300K if 1-2 hospitals are connected and 1-2 pharma/CRO partnerships exist.

Challenges

• Regulatory complexity: EHDS secondary use framework isn't finalized. GDPR consent mechanisms for data intermediaries are complex. Each country has different rules.
• Hospital willingness: Hospitals are nervous about "selling patient data." Even with consent and de-identification, there's reputational risk. The framing matters: "enabling clinical trials" not "selling data."
• Chicken-and-egg: Pharma won't pay until there's data volume. Hospitals won't invest in curation until there's pharma demand. Need a lighthouse hospital + a lighthouse pharma partner simultaneously.
• De-identification: Must be provably robust. GDPR fines for re-identification are massive. This is a trust liability.

Verdict

High potential, high difficulty. The revenue per transaction is high, but the regulatory and trust barriers are significant. Best pursued as a year 2-3 opportunity after establishing the core EHDS compliance product. Start with the hospital relationship first, add the intermediary layer later.

2. B2B Preventive Health — Employee Health Programs

Concept: Employer pays for employee health screening, data is stored in a PHKG, tracked over years. Employees get personalized health insights. Employer gets healthier workforce, reduced sick days, insurance premium optimization.

How It Works

1. Employer contracts Stamen for employee health program
2. Employees get periodic health screenings (blood tests, body scans, questionnaires) — can partner with Neko Health, Nightingale Health, or local providers
3. Screening data is curated into individual PHKGs
4. AI analyzes trends over time: cholesterol trajectory, blood pressure trends, metabolic markers
5. Employee gets personalized health dashboard with recommendations
6. Employer gets anonymized aggregate workforce health reports

Market Context

Corporate wellness market: Estimated $61B globally (2024), growing 7-8% annually. Nordic corporate wellness is a mature market with strong employer mandates.

Existing players:

• Neko Health (Sweden, $2.55B) — full-body scanning, currently B2C. No PHKG or longitudinal tracking.
• Nightingale Health (Finland, public) — blood test biomarker platform. Sells to employers for health screening.
• Wellhub/Gympass (Brazil/US) — fitness/wellness platform for employees. Not health data.
• Virgin Pulse / Personify Health — enterprise wellness platforms. Generic, not data-centric.

What's different:

• Existing corporate wellness is gym memberships, mental health apps, generic health tips. Nobody does longitudinal health data tracking with a PHKG.
• The PHKG approach means data accumulates over years. An employee's 5-year health trajectory is far more valuable than a single screening.
• AI can identify early warning signs across multiple data points that no single test catches.

Revenue Potential

• Per-employee subscription: €20-50/month per employee for screening + data platform
• Employer value prop: Reduced sick days (typical ROI: €2-5 saved per €1 spent on wellness)
• Example: 500 employees × €30/month = €15,000/month = €180,000/year
• With 10 employer clients: €1.8M ARR

Challenges

• Employee privacy concerns: Employer-funded health data creates privacy tensions. Employees worry about data being used for employment decisions. Must have iron-clad data separation (employee sees their data, employer only sees anonymized aggregates).
• Not a core competency: This is a healthcare product, not a data infrastructure product. Different sales motion, different buyer (HR director, not hospital CIO). Distracts from the core EHDS play.
• Screening provider dependency: Stamen would need to partner with screening providers (Neko, Nightingale). Margins get squeezed.
• Adoption: Employee participation rates in corporate wellness programs average 20-40%. Getting 50%+ participation requires strong employer mandates.

Verdict

Good revenue potential, but a distraction from core EHDS play. Best pursued as a later product line (year 3+) or through a partnership (e.g., provide the PHKG backend for Neko Health's B2B offering). Don't build the screening/sales operation yourself — partner with someone who already has employer relationships.

3. Patient Group Data Consolidation Apps

Concept: Patients with complex or chronic conditions (rare diseases, cancer, autoimmune) have data scattered across 5-20 hospitals, labs, and specialists. The app consolidates everything into a PHKG, AI helps curate and explain it, patients can use it for second opinions.

How It Works

1. Patient downloads app, connects to hospital portals / uploads records
2. Data is ingested, structured into PHKG using SNOMED CT
3. AI generates a unified health timeline and summary
4. Patient can ask questions about their data ("What did my MRI in 2023 show about the lesion?")
5. App highlights gaps, contradictions, or things to discuss with next doctor
6. Patient can share curated PHKG with new specialists for second opinions

Market Context

Existing players:

• PicnicHealth (US, $60M+ raised) — collects records from US hospitals, provides patient access. 12 of top 20 pharma customers. But: no AI curation, no knowledge graph, US-only.
• 1upHealth (US, $40M) — FHIR-based patient data aggregation. CMS-compliant. But: no AI, no PHKG.
• Citizen Health (US) — rare disease patient data platform. Building AI infrastructure for patient data. Early stage.
• Betterpath (US) — patient record consolidation. Very early.

What's different:

• PicnicHealth collects but doesn't curate or analyze. It's a record storage service.
• PHKG approach means the data is actually structured, queryable, and AI can reason over it.
• The "talk to your data" feature (AI explaining your records) is unique. Nobody does this well today.
• Second opinion facilitation — sharing a curated PHKG with a new doctor is more valuable than sending 200 PDFs.

Revenue Potential

• B2C subscription: €10-30/month per patient
• Target population: Rare disease patients (30M in EU), cancer survivors, autoimmune patients
• Niche first: Start with ONE rare disease community (e.g., Ehlers-Danlos, ME/CFS) where patients are highly motivated and vocal
• Example: 5,000 patients × €15/month = €75,000/month = €900,000/year

Challenges

• B2C is hard in healthcare: Patients don't pay for health data tools. Adoption is driven by free value, not subscriptions. The freemium-to-paid conversion in health apps is 1-5%.
• Hospital portal integration: Getting data out of hospital EHR systems is technically and legally complex. Each hospital has different portals, different data formats, different access rules.
• Patient motivation: Most patients don't care about their data until they need a second opinion. The app needs a trigger event (new diagnosis, hospital change, specialist visit) to drive adoption.
• Data quality: Hospital records are messy. Auto-curation needs to handle missing data, contradictory notes, outdated information.

Verdict

Compelling product, hard business model. B2C health apps rarely make money from subscriptions. But: this is the fastest to prototype (can build in 3-6 months), and the community/word-of-mouth potential is high if you pick the right patient group. Best pursued as a proof-of-concept that validates the PHKG technology with real users, even if revenue is modest initially. The real value is in demonstrating the PHKG works with messy real-world data.

4. GP Diagnostic Overview — AI Summarisation Layer on Kjernejournal

Concept: GP has a patient in front of them. The PHKG provides an AI-summarised, structured, ontology-backed view of the patient's history — not just a flat document list. AI highlights relevant history for the current complaint. GP diagnoses faster with fewer redundant tests.

Important positioning: This is NOT "first-ever cross-institutional view for Norwegian GPs." Norway already has Kjernejournal (national Summary Care Record, since 2017) — accessible from inside the GP's EHR, including overview of clinical documents from all hospitals (discharge summaries, x-ray results, lab results) via IHE-XDS protocol. Pasientens Legemiddelliste (PLL) is rolling out from 2024, providing national patient medication access. The product is an AI summarisation layer on top of Kjernejournal/PLL, not a replacement for existing infrastructure.

How It Works

1. Patient data flows through Kjernejournal (existing national infrastructure)
2. Stamen's PHKG ingests Kjernejournal documents + unstructured clinical notes
3. AI structures, codes (SNOMED CT), and summarises the full history
4. GP sees a structured timeline + AI summary, not a flat document list
5. AI highlights: relevant past conditions, current medications, trends, risk factors
6. GP can ask natural language questions: "Has this patient ever had a cardiac workup?"

The Underlying Problem — With Sources

• History dominates diagnosis: Hampton et al. (1975) found history alone yields the correct diagnosis in 75-80% of cases. Replicated by Peterson et al. (1992, 76%) and Roshan & Rao (2000, 83%). If the GP has a better-structured history, consultation time shifts to treatment.

• Time pressure is severe: Wilson et al. (2024, BMJ Open) found delivering recommended care would require >24 hours per GP per day.

• Kjernejournal is flat: It provides a document store — discharge summaries, lab results, imaging reports — but not structured, ontology-backed, AI-summarised views. GPs still need to manually read and cross-reference multiple documents. The gap is intelligent summarisation, not access.

• Norwegian GP consultation time: No verified primary source found for a single average figure.

Clinical Safety — The Real Constraint

The evidence for AI chart summarisation is promising but has serious safety signals:

The promise:

• ChatGPT-4 adapted with in-context learning performed superior to physicians on both AI metrics and clinical expert evaluations.
• Best-in-class systems achieve 1.47% hallucination rate and 3.45% omission rate across 12,999 clinician-annotated sentences.

The risks:

• Omissions are worse than hallucinations: A 2026 study of an EHR-integrated AI chart review tool deployed nationally in the US found that missing/confusing information was more commonly reported than hallucinations. Errors of omission may be a larger threat than errors of commission.
• 47% omission rate in ED summaries: A separate evaluation found 47% of 100 LLM-generated emergency department summaries omitted clinically relevant information.
• Automation bias: A 2025 study of Epic's GPT-4 summarisation tool at NYU Langone found 22.7% of providers reported sometimes skipping full-length notes in favor of the summary alone. If the summary omits a relevant finding, the tool causes the diagnostic miss it was supposed to prevent.
• Verification overhead erases time savings: PDSQI-9 validation required clinician evaluators averaging 10 minutes per evaluation. If GPs need to verify the summary against source notes, the time saving disappears.

Implication: The clinical safety case (omissions, automation bias, verification cost) is the binding constraint — not technology readiness or integration cost.

Competitive Landscape

Major competitors (wiki previously omitted):

• Epic's own AI chart summarisation (GPT-4 powered): Deployed nationally across US health systems including NYU Langone, scaled in 2024-25. This is the product Stamen would compete with directly. Source: NYU Langone study on automation bias.

• Helseplattformen (Norwegian Epic deployment): Mid-Norway region. Adds complexity — potentially gives Epic a foothold for chart-summarisation features in Norway. Stamen would be competing against Epic's own AI features in Helseplattformen regions.

• Abridge, Nabla, Suki, DeepScribe: Primarily ambient documentation but expanding into chart review/summarisation. Well-funded.

• Microsoft/Nuance DAX Copilot: Same direction — ambient documentation expanding to chart summarisation.

• Tandem Health (Sweden/Norway) — AI clinical documentation with EHR access already negotiated. Nordic competitor with market presence.

• Cogstack (UK) — NHS clinical NLP, open source, extracts from unstructured notes. Research tool, not commercial product.

The framing "no existing GP tool does AI summarisation of full history" was true in 2023. By 2026 it is actively contested.

Norwegian GP EHR Landscape (corrected)

Norwegian GP EHRs are primarily CGM Allmenn (formerly WinMed/Profdoc), Infodoc Plenario, and Pridok. DIPS is the dominant hospital EHR, not a GP EHR. Getting this wrong undermines credibility with Norwegian GP buyers.

Revenue Potential

• Per-practice subscription: €200-500/month per GP practice
• Norwegian GP market: ~5,000 GPs, ~2,500 practices
• Example: 100 practices × €300/month = €30,000/month = €360,000/year
• But: Epic's summarisation comes "free" with the EHR in Helseplattformen regions. Pricing pressure is real.

Challenges

• Clinical safety is the binding constraint: Not integration cost. Not technology readiness. A prospective clinical study showing the summarisation doesn't increase diagnostic misses is a prerequisite for GP trust and regulatory clearance.
• MDR / AI Act compliance: Any tool that highlights "relevant" history for the current complaint makes a clinical relevance judgment → clinical decision support under EU MDR → likely 12-18 months and significant capital for CE marking. Klinik has been navigating this for a decade; Stamen starts from scratch.
• Epic competition: In Helseplattformen regions, Epic's own GPT-4 summarisation comes bundled. Stamen needs to be demonstrably better, not just different.
• Integration: Connecting to CGM Allmenn, Infodoc Plenario, Pridok is technically complex. Each has different APIs, data models, authentication. 6-12 months per EHR.
• Automation bias: If GPs start relying on summaries without verifying, the tool becomes a liability. Product design must include verification nudges.

Verdict

The problem is real, but the opportunity is harder and narrower than it first appears.

• Norway already has Kjernejournal and PLL — the product is an AI summarisation layer on top, not first-ever cross-institutional access.
• The defensible niche may be narrower than "GP diagnostic overview": something like "structured ontology-backed summary for complex multimorbidity patients where omission risk is verifiable against the underlying PHKG" — leveraging the actual differentiator (PHKG structure) rather than competing on generic summarisation.
• Gating item is regulatory clearance and a published clinical safety evaluation, not just data flow. Budget for a prospective clinical study, not just integration work.
• Epic's own summarisation tool, bundled with Helseplattformen, is the direct competitor. Stamen must be provably better on accuracy/omission rates, not just different on architecture.
• Year 2-3 timing is plausible but depends on completing clinical safety validation. Not before.

5. Pre-Consultation Triage — Patient Answers Before the Visit

Concept: Before a GP consultation, the patient answers structured questions via an app. AI triages: what symptoms to focus on, what tests to pre-order, what the GP should prioritize. Saves consultation time, improves outcomes.

How It Works

1. Patient books GP appointment
2. 24-48 hours before: patient gets a questionnaire (structured questions based on reason for visit)
3. Patient answers: symptoms, duration, severity, relevant history
4. AI triages: suggests tests to pre-order (blood work, imaging), flags urgent cases
5. GP receives summary before the consultation: patient's answers + AI triage + relevant PHKG history
6. Consultation focuses on diagnosis and treatment, not data gathering

Market Context

Existing players:

Klinik Healthcare Solutions (Finland, founded 2013) — the serious incumbent. CE-marked, deployed across NHS GP practices in England. AI patient flow management that routes patients to appropriate care. Independent health-economic evaluation by York Health Economics Consortium at Priory Medical Group: enabled delivery of 18 months' appointments in one year, £300k+ capacity savings. However: Finnish entity reported €1.49M revenue and €701k loss in FY2023, revenue down 6.1%, headcount down 14.3%. UK revenue likely booked separately. This is a signal about category willingness-to-pay, not just execution.

NHS England incumbents: The UK is not a greenfield. AccuRx, eConsult, Patchs, and Anima are already widely deployed across NHS GP practices. The NHS Long Term Plan effectively mandated electronic triage forms. These represent the realistic competitive set if Stamen targets UK or follows UK-style procurement in the Nordics.

Ada Health (Germany, $100M+ raised) — AI symptom assessment with 30K+ medical concepts. B2C app, not integrated into GP workflow. Different product (patient-facing diagnosis vs. GP-assistant triage).

Babylon Health (UK, $4B peak valuation → collapsed) — AI triage + telehealth. Failed due to overpromising and regulatory issues. Warning story: don't position as "AI doctor."

Your.MD (Norway/UK) — AI symptom checker. Pivoted, limited traction.

What's different:

• Ada and Babylon are standalone symptom checkers. They compete with the GP, not assist the GP.
• Klinik and NHS incumbents do static questionnaire triage. A PHKG-backed version with LLM-based structured history-taking is a different product.
• The LLM-augmented version is genuinely new and promising. The static-questionnaire version has a 25-year history of underwhelming results (see RCT evidence below).

RCT Evidence — What Actually Works

Static questionnaire RCTs (1997-2024): mostly null or modest results.

• ESOGER trial (2024): null results on consultation length and outcomes.
• VISIT trial (2022): no satisfaction effect, slight agenda-setting increase.
• These trials tested static forms filled before appointments. The evidence is clear: static questionnaires alone don't meaningfully reduce consultation time or improve outcomes.

LLM-augmented version: genuinely promising.

• PreA RCT (Nature Medicine, January 2026, n=2,069, 111 specialists): 28.7% reduction in consultation duration using LLM-based structured history-taking. However: this was specialist care, not GP. Generalisation to primary care is uncertain.

Key insight: The defensible version of this opportunity is narrower than generic "pre-consultation triage": LLM-based structured history-taking integrated into the GP's existing EHR, positioned as assistant not replacement, with a measurable consultation-time outcome.

Revenue Potential

• Per-consultation fee: €2-5 per pre-consultation triage
• Or per-practice subscription: €100-300/month per GP practice
• Norwegian GP consultations: ~20 million/year across ~5,000 GPs
• Example: If used for 10% of consultations at €3 each = 2M consultations × €3 = €6M addressable market in Norway alone

Challenges

• Simplicity is everything: If the questionnaire takes more than 3 minutes, patients won't do it. Must be dead simple.
• GP trust: GPs will ignore AI triage suggestions if they're wrong 20% of the time. Need very high accuracy for the specific triage use case (not general diagnosis).
• Babylon warning: Babylon raised $4B and collapsed. The lesson: AI triage in healthcare is harder than it looks, and overpromising kills trust. Must underpromise and overdeliver.
• Test pre-ordering: Getting labs to accept pre-orders from an app requires integration with lab systems. Not trivial.
• Regulatory: If the tool suggests specific tests, it's clinical decision support → potentially needs MDR certification.

Verdict

The problem is real and well-evidenced, but "simplest product, fastest to revenue" is misleading. The simplicity is in the build, not in trust, adoption, regulatory clearance (MDR), or willingness-to-pay. A decade-old, CE-marked, NHS-deployed Nordic competitor (Klinik) doing ~€1.5M revenue and losing money is the relevant base rate.

The defensible version is narrower: LLM-based structured history-taking integrated into the GP's existing EHR, positioned as assistant not replacement, with a measurable consultation-time outcome. The PreA RCT (28.7% reduction) is the evidence to build on. Static questionnaires have 25 years of underwhelming RCT results — don't build that.

Regulatory risk: If the tool suggests specific tests or diagnoses, it's clinical decision support → potentially needs CE marking as a medical device (MDR). Questionnaire-only approach has lower regulatory burden but also lower value.

Babylon is a warning: Don't position as "AI doctor." Position as "structured history-taking assistant." Underpromise, overdeliver."

Recommended Sequencing

Phase	Product	Why This Order
Year 1	5. Pre-Consultation Triage	Simplest product, fastest to revenue, validates GP relationship
Year 1-2	3. Patient Group Apps	Proves PHKG with real users, builds community, generates case studies
Year 2	1. EHDS Compliance (core)	Core product — hospital data curation for EHDS. Builds the data foundation.
Year 2-3	4. GP Diagnostic Overview	Uses PHKG data from EHDS product. High value, high integration cost.
Year 3+	2. B2B Preventive Health	Partnership play. Don't build screening/sales operation.
Year 3+	1. Data Intermediary	Add layer after hospital data is flowing. High regulatory complexity.

Start with the simplest product that proves the technology and generates revenue. Build toward the core EHDS play. Add data monetization layers only after the data infrastructure is established.

See Also

• Stamen Health — strategic positioning
• Stamen Health Executive Summary — one-page plan
• AIDAVA — research foundation
• PHKG Business Models & Market — market sizing
• Companies — competitor database
• AIDAVA Competitive Analysis — detailed competitive landscape