💰 Funding Details

End user-driven application of Generative Artificial Intelligence models in healthcare (GenAI4EU)

Overview

Horizon Europe Cluster 1 – Health has launched the single-stage call HORIZON-HLTH-2025-01-CARE-01 for Research & Innovation Actions (RIA). The call funds large-scale, multinational consortia that will design, develop and clinically validate trustworthy, explainable virtual assistant solutions based on Generative AI – augmented by other AI tools – to support healthcare professionals across the care continuum.

* Opening date: 22 May 2025

* Deadline: 16 September 2025, 17:00 (CET)

* Indicative budget per project: up to €80 million

* Maximum EU co-funding rate: 100 % of eligible costs

* Type of MGA: HORIZON Action Grant (budget-based)

* Destination: *Ensuring equal access to innovative, sustainable, and high-quality healthcare*

Expected Outcomes

1. Clinician-centred virtual assistants that are robust, user-friendly and aligned with EU trustworthy-AI principles.

2. Cross-border methodologies fostering acceptance, uptake and public trust.

3. Improved patient outcomes through personalised care and stronger patient–provider communication.

4. Health-system efficiencies via superior accuracy, safety and cost-effectiveness compared with standard of care.

Scope Highlights

- Develop new or optimised GenAI models trained on multimodal, multi-lingual and federated datasets (EHR, imaging, omics, unstructured notes, etc.).

- Demonstrate clinical utility in ≥ 2 use-cases covering different medical fields and unmet needs.

- Co-create with healthcare professionals, patients and SSH experts; integrate training programmes for end-users.

- Design a regulatory & HTA evidence plan, covering CE-marking, AI Act, MDR and liability aspects.

- Implement continuous assessment frameworks for bias mitigation, explainability, ELSI and GDPR-compliance.

- EU added value: leverage EU-developed models, AI Factories, EOSC services, European Health Data Space, etc.

Eligibility Snapshot

- Minimum consortium: 3 independent legal entities from 3 different EU/Associated your country.

- Involvement of industry (incl. start-ups), academia, hospitals, patient organisations, SSH experts and, optionally, the JRC is strongly recommended.

- Entities from the USA may be funded; high-risk 5G suppliers are excluded.

Evaluation Weights (thresholds)

*Excellence* 4/5 *Impact* 4/5 *Quality & efficiency of implementation* 4/5 (cumulative ≥ 12).

📊 Evaluation Criteria

Horizon Europe scoring focus (thresholds: 4/4/4)

Excellence

* Innovation beyond state-of-the-art GenAI, multimodal & multilingual.

* Sound methodology for bias mitigation & explainability.

Impact

* Credible pathway to healthcare cost-effectiveness and equitable access.

* Dissemination & exploitation plan, including open-source components.

Implementation

* Complementary consortium, clear KPIs, risk management, realistic budget.

📝 Application Strategy

Strategic Application Guide

1. Build a Mission-driven Consortium

1. Clinical leadership – Engage reference hospitals for each targeted use-case (e.g. oncology, intensive care).

2. AI & data science excellence – Secure partners with proven GenAI model-development track records.

3. Data holders – Formalise data access to *national EHDS pilots, biobanks, imaging repositories*.

4. SSH & ethics – Include legal scholars, ethicists and health economists early.

5. SME/start-up participation – Demonstrate agility and pathways to market.

2. Craft a Convincing Excellence Narrative

* Novelty – Explain how your GenAI architecture surpasses LLM baselines (e.g. hybrid neuro-symbolic models, multimodal fusion).

* Federated Learning – Detail privacy-preserving training across your country nodes.

* Clinical research plan – Provide robust PICO, statistical power and TRL targets (aim ≥ TRL 6 by project end).

3. Maximise Impact

* Map outcomes to EU policies (AI Act, EHDS, Cancer Mission, Rare Disease Partnership).

* Quantify health-economic gains (cost-utility analyses vs. standard of care).

* Outline exploitation routes – open-source core models; proprietary clinical wrappers (dual-licensing).

* Design a scalability blueprint for deployment across diverse healthcare systems.

4. Implementation & Risk Management

* Work-package logic – WP1 Management; WP2 Data Governance; WP3 Model Dev; WP4 Clinical Pilots; WP5 Evaluation & HTA; WP6 Ethics & SSH; WP7 Dissemination & Exploitation.

* GANTT & PERT – Show interdependencies, buffer for ethics approvals.

* Risk register – Model hallucinations, data-access delays, regulatory changes; provide mitigations.

5. Budgeting Tips

* Allocate ≥ 5 % for joint clustering activities with other projects.

* Ring-fence resources for patient involvement stipends and professional translation (multi-lingual datasets).

* Plan for cloud/HPC costs via AI Factories or EuroHPC vouchers.

6. Submission Quality Control

* Use proposal checklists aligned with HEU Part B; respect 45-page limit.

* Conduct red-team review (mock evaluators) two weeks before deadline.

* Submit at least 24 h early to avoid IT bottlenecks.

💡 Success Factors

Early Regulatory Dialogue

Establish **pre-submission meetings with notified bodies and HTA agencies** within the first six months to align evidence generation with MDR and AI Act expectations.

Federated Learning Infrastructure in Place Day-1

Deploy or partner with an existing **GAIA-X compliant federated platform** (e.g. TEF-Health node) to fast-track multi-your country data training without waiting for new IT builds.

Co-design Sprints with Clinicians & Patients

Run **quarterly design thinking workshops** where healthcare professionals and patient advocates iteratively test prototypes, ensuring usability and trustworthiness.

Robust Explainability Layer

Integrate **post-hoc (SHAP, LIME) and intrinsic (attention maps)** explanation modules, and validate them through **user comprehension studies** documented in deliverables.

Dedicated Health-Economic Evaluation Work-Package

Quantify **incremental cost-effectiveness ratios (ICERs)** and budget-impact models early to demonstrate health-system value, bolstering the Impact criterion.

Synergy with EU Infrastructures

Explicitly leverage **European Health Data Space pilots, EOSC services and AI Factories**, showing how resources reduce duplication and amplify EU added value.

❌ Common Mistakes to Avoid

Underestimating Data Governance Complexity

Applicants frequently overlook the time and legal work needed to negotiate **cross-border data-sharing agreements** that satisfy GDPR, EHDS and hospital policies. Vague statements like “data will be available” without MOUs or DTA drafts score poorly.

Weak Clinical Evidence Plan

Proposals sometimes present ambitious GenAI prototypes but lack a **clear clinical study design** (end-points, control arm, sample size). Evaluators penalise projects that cannot demonstrate how superiority over standard of care will be statistically proven.

Ignoring SSH and Patient Engagement

Leaving social scientists or patient organisations as peripheral partners (or omitting them) is a critical error. The call explicitly demands **multidisciplinary, end-user-driven development**.

Superficial Bias Mitigation Strategy

Simply stating that fairness libraries will be used is insufficient. Evaluators expect **concrete bias audits** (e.g. subgroup performance metrics by sex, age, ethnicity) and iterative mitigation loops.

One-size-fits-all Budget

Flat percentage allocations across partners signal poor planning. Each WP should have a **justified, activity-linked cost breakdown** (e.g. HPC credits, annotation tasks, clinician time).

Missing Alignment with Upcoming AI Act

Forgetting to reference **high-risk AI system requirements** and conformity assessment pathways under the EU AI Act can raise doubts about market-readiness.

End user-driven application of Generative Artificial Intelligence models in healthcare (GenAI4EU)

Quick Facts for Spain Applicants

Ready to Apply from Spain?

💰 Funding Details

End user-driven application of Generative Artificial Intelligence models in healthcare (GenAI4EU)

Overview

Expected Outcomes

Scope Highlights

Eligibility Snapshot

Evaluation Weights (thresholds)

📊 Evaluation Criteria

Horizon Europe scoring focus (thresholds: 4/4/4)

Excellence

Impact

Implementation

📝 Application Strategy

Strategic Application Guide

1. Build a Mission-driven Consortium

2. Craft a Convincing Excellence Narrative

3. Maximise Impact

4. Implementation & Risk Management

5. Budgeting Tips

6. Submission Quality Control

💡 Success Factors

Early Regulatory Dialogue

Federated Learning Infrastructure in Place Day-1

Co-design Sprints with Clinicians & Patients

Robust Explainability Layer

Dedicated Health-Economic Evaluation Work-Package

Synergy with EU Infrastructures

❌ Common Mistakes to Avoid

Underestimating Data Governance Complexity

Weak Clinical Evidence Plan

Ignoring SSH and Patient Engagement

Superficial Bias Mitigation Strategy

One-size-fits-all Budget

Missing Alignment with Upcoming AI Act

🇪🇸Spain Overview

🏷️ Keywords

🔗 Quick Links

Related Opportunities for Spain

Leveraging artificial intelligence for pandemic preparedness and response

Leveraging multimodal data to advance Generative Artificial Intelligence applicability in biomedical research (GenAI4EU)

Using Generative AI (GenAI4EU ) for Scientific Research via EOSC

Ready to Apply from Spain?

Frequently Asked Questions