FeatureCloud · Project

Secure AI Platform That Analyzes Patient Data Without Moving It Out of Hospitals

healthPilotedTRL 7

Imagine ten hospitals want to train an AI to detect cancer — but none of them can legally share patient files with each other. FeatureCloud built a system where each hospital trains the AI locally and only shares the lessons learned, never the actual records. It's like each chef tasting their own soup and sharing only the recipe improvements, so nobody's secret ingredients ever leave the kitchen. On top of that, blockchain technology gives patients a real "off switch" — they can revoke consent and their data is truly removed.

By the numbers

EUR 4,646,000

EU funding for development

consortium partners

countries in cross-border testing

SMEs in the consortium

total project deliverables

demo deliverables with working software

30%

industry ratio in consortium

The business problem

What needed solving

Hospitals and health organizations sitting on valuable patient data cannot pool it for AI training because of privacy laws, cyber risks, and patient trust concerns. This creates a deadlock: the AI needs data from many hospitals to be accurate, but no hospital can legally or safely share it. Meanwhile, every data exchange point is a potential target for cyberattacks.

The solution

What was built

FeatureCloud built a complete federated machine learning platform with an app store for deploying AI models across distributed hospital sites, blockchain-based patient consent management, downloadable federated clustering and multi-OMICS analysis software, and layman-friendly interfaces for patients, project managers, and developers.

Audience

Who needs this

Multi-site hospital groups wanting to train AI across locations without moving patient dataPharmaceutical companies running multi-center clinical trials needing distributed data analysisHealth IT vendors building GDPR-compliant analytics platformsNational health services planning cross-border data collaborationMedical device companies needing certifiable AI tools for diagnostics

Business applications

Who can put this to work

Hospital Networks & Health IT

enterprise

Target: Multi-site hospital groups or health IT providers managing clinical data across locations

If you are a hospital network dealing with the impossibility of pooling patient data across sites for AI diagnostics — this project developed a federated machine learning platform with an app store that lets you train AI models across all your locations without any sensitive data ever leaving each hospital. The platform was built by 10 partners across 5 countries and includes blockchain-based consent management so patients can revoke access at any time.

Pharmaceutical & Clinical Research

enterprise

Target: Pharma companies or CROs running multi-center clinical trials

If you are a pharma company struggling to aggregate real-world patient data from multiple trial sites for drug discovery — this project built downloadable federated clustering and multi-OMICS analysis software that works across distributed datasets. With EUR 4,646,000 in EU funding and 41 deliverables over 5 years, the toolkit was designed for certifiable medical device standards, meaning it can fit into regulated workflows.

Health Data Platforms & Cybersecurity

mid-size

Target: Health data exchanges, insurance tech firms, or cybersecurity vendors serving healthcare

If you are a health data platform provider worried about cyber risks to your infrastructure and GDPR compliance — this project created a privacy-by-architecture cloud platform where no sensitive data travels through communication channels and there is no central point of attack. The system includes layman-friendly interfaces usable by patients, project managers, and developers, lowering the barrier to adoption.

Frequently asked

Quick answers

What does FeatureCloud's platform actually cost to deploy?

The project was funded with EUR 4,646,000 in EU contribution as a Research and Innovation Action. Pricing for commercial use is not specified in the project data. Contact the University of Hamburg or check featurecloud.eu for current licensing or deployment options.

Can this scale beyond a research pilot to real hospital networks?

The deliverables include a working app store with federated machine learning apps, downloadable clustering pipeline software, and interfaces designed for patients, project managers, and developers — all signs of production-readiness. The consortium of 10 partners across 5 countries (AT, DE, DK, NL, RO) tested the system in a distributed, cross-border setup.

What is the IP situation — can we license this technology?

FeatureCloud was funded as a Research and Innovation Action (RIA) under Horizon 2020, which typically means IP stays with the consortium partners. The University of Hamburg coordinated; 3 industry partners and 3 SMEs were involved. Specific licensing terms should be negotiated directly with the consortium.

Does this meet healthcare data regulations like GDPR?

The entire platform was designed around privacy-by-architecture: no sensitive data leaves the local site, there is no central data storage point, and blockchain gives patients the ability to revoke consent at any time. The project explicitly addressed legal considerations, international policies, and certification as medical devices.

How long would it take to integrate this into our existing hospital IT?

The project ran from 2019 to 2023 and delivered layman-friendly interfaces, an extendible app store, and downloadable software packages. Based on the deliverable descriptions, the toolkit was designed for integration rather than replacement. Exact integration timelines depend on your infrastructure.

What kind of AI applications can run on this platform?

The app store supports federated machine learning applications including multi-OMICS system-medicine clustering and general federated data mining. The platform is extendible by developers, meaning custom AI models can be added. All 8 demo deliverables confirm working software rather than conceptual designs.

Is there ongoing support or is this a finished research project?

The project officially closed in December 2023. However, software is available for download and the app store was built to be extendible. Check featurecloud.eu for current status, community activity, and whether commercial support is offered by consortium members.

Consortium

Who built it

The FeatureCloud consortium is well-balanced for taking research to market: 10 partners across 5 European countries (Austria, Germany, Denmark, Netherlands, Romania), with 6 universities providing deep AI and medical expertise, 3 SMEs bringing commercial agility, and 1 research organization. The 30% industry ratio is solid for a Research and Innovation Action — it means real companies were at the table shaping the product, not just academics building for other academics. The coordinator, University of Hamburg, anchors the project in Germany's strong health-tech ecosystem. The cross-border setup across 5 countries also means the system was tested against different healthcare regulations and IT infrastructures, which is a genuine advantage for any buyer considering multi-country deployment.

UNIVERSITY OF HAMBURGCoordinator · DE
PHILIPPS UNIVERSITAET MARBURGparticipant · DE
MEDIZINISCHE UNIVERSITAT GRAZparticipant · AT
TECHNISCHE UNIVERSITAET MUENCHENparticipant · DE
UNIVERSITEIT MAASTRICHTparticipant · NL
SYDDANSK UNIVERSITETparticipant · DK
SBA RESEARCH GEMEINNUTZIGE GMBHparticipant · AT
CONCENTRIS RESEARCH MANAGEMENT GMBHparticipant · DE
RESEARCH INSTITUTE AG & CO KGparticipant · AT

How to reach the team

University of Hamburg (Germany) — reach out to the FeatureCloud project lead via the university's computer science or medical informatics department

Order a report on this consortium See UNIVERSITY OF HAMBURG profile →

Next steps

Talk to the team behind this work.

Want an introduction to the FeatureCloud team? SciTransfer can connect you with the right people and provide a detailed technology brief tailored to your use case.

Order a report on this topic More in Health & Biomedical