Artificial intelligence is opening new possibilities for healthcare, but developing reliable AI models often requires access to large and diverse datasets. In healthcare, this creates a fundamental challenge: patient data is highly sensitive, subject to strict regulatory requirements and typically distributed across hospitals and research institutions.
The FLUTE (Federated Learning and mUlti-party computation Techniques for prostatE cancer) project explored how organisations can collaborate on healthcare AI without centralising or transferring sensitive patient data. For Technovative Solutions Ltd. (TVS), the project provided an opportunity to develop advanced federated healthcare infrastructure and contribute to the practical application of privacy-enhancing technologies.
As the project concludes, we look back at what FLUTE set out to achieve, the technologies developed, TVS's contribution and the potential of this work beyond prostate cancer.
What Was FLUTE and What Did It Aim to Achieve?
FLUTE was one of the Horizon Europe Research and innovation projects focused on developing privacy-preserving technologies for healthcare AI, with prostate cancer as its primary clinical use case.
The project addressed a key challenge in modern healthcare research: how to make use of distributed clinical data for AI development while ensuring that sensitive patient information remains protected.
Rather than requiring hospitals and research institutions to pool their raw data in a central repository, FLUTE explored a federated approach in which AI models can be trained across distributed data sources while the underlying patient data remains within the participating organisations.
For prostate cancer, the project aimed to demonstrate how this approach could support the development and validation of AI models for predicting clinically significant disease. More broadly, FLUTE sought to contribute to a scalable framework for privacy-preserving healthcare research and AI innovation.
How Did FLUTE Protect Patient Data?
Federated Learning (FL) was at the centre of the FLUTE approach. Instead of moving patient data to a central location, models are sent to participating data hubs, where training takes place locally. Only protected information related to the model is exchanged.
FLUTE strengthened this architecture through a range of Privacy Enhancing Technologies (PETs), including:
- Secure Multi-Party Computation (SMPC)
- Homomorphic Encryption
- Differential Privacy
- Trusted Execution Environments (TEEs)
These technologies provide additional layers of protection during computation, model training and information exchange.
The approach is particularly relevant to healthcare because it enables institutions to participate in collaborative AI research while retaining control over their sensitive datasets. It also supports the principles of data minimisation, security and controlled data access that are central to GDPR-compliant healthcare data management.
FLUTE also addressed interoperability through technologies and standards such as Fast Healthcare Interoperability Resources (FHIR) and Digital Imaging and Communications in Medicine (DICOM), helping connect heterogeneous clinical and medical imaging data sources within the federated environment.
What Was TVS's Role in FLUTE?
TVS played a central technical role in the development of the FLUTE platform.
The company led the development of the platform within the project, including the data hub and innovator dashboards, as well as work related to scalability, usability, performance validation, privacy protection validation and platform documentation.
The TVS team also contributed to the integration of different technical components developed across the consortium. This included backend services, APIs, data harmonisation and interoperability mechanisms, helping bring the project's different technologies together within a usable federated healthcare environment.
The platform was designed around two principal stakeholder groups:
Data Hub Owners, such as hospitals and healthcare organisations, retain control over their data and manage how it can be used.
Researchers and Innovators can discover available datasets, define research requirements, create cohorts, develop AI models and conduct federated research without directly accessing or transferring sensitive patient information.
This work drew on TVS's broader expertise in software development, data analytics, machine learning and AI, secure digital infrastructure and complex systems integration.
FLUTE Project’s Key Outputs and Achievements: TVS’ contribution
One of the most important outcomes of FLUTE was demonstrating how Federated Learning and advanced Privacy Enhancing Technologies can be brought together within a practical healthcare platform.
The project developed an ecosystem connecting distributed healthcare data hubs with researchers and AI workflows, while incorporating security, privacy, interoperability and governance considerations.
For TVS, an important achievement was advancing the FLUTE platform from individual technical components towards an integrated environment. The work included developing dashboards and services, supporting healthcare data interoperability, integrating partner technologies and contributing to validation of platform performance and privacy.
The project also built on knowledge and components from the TRUMPET Project, enabling TVS to further develop its expertise in federated healthcare AI and privacy-preserving technologies.
These achievements are significant because they demonstrate that protecting patient data does not necessarily mean limiting collaborative AI research. With the appropriate architecture and safeguards, distributed healthcare data can contribute to collaborative innovation while remaining under the control of its owners.
Where Can These Technologies Be Applied Beyond Prostate Cancer?
Although prostate cancer was the primary FLUTE use case, the underlying technologies have much broader potential.
The same federated and privacy-preserving approach can support research involving other cancers, medical imaging, rare diseases, cardiovascular conditions, neurological disorders and public health.
TVS's experience also demonstrates the potential of multimodal healthcare AI, where structured clinical information can be combined with medical imaging and other data sources.
Beyond individual clinical studies, federated architectures can support wider healthcare ecosystems in which hospitals, research organisations and technology providers need to collaborate without creating a centralised repository of sensitive patient information.
What Does FLUTE Mean for TVS's Future?
FLUTE has strengthened TVS's capabilities across several areas that are becoming increasingly important in digital healthcare: Federated Learning, Privacy Enhancing Technologies, healthcare interoperability, multimodal data, secure data platforms and AI-enabled research workflows.
The project has also provided practical experience in integrating complex technologies into a functioning healthcare ecosystem, while addressing privacy, security, governance and interoperability requirements together rather than in isolation.
This expertise builds on TVS's broader work in digital healthcare and supports its vision of creating secure, interoperable and privacy-preserving healthcare ecosystems.
As healthcare AI continues to evolve, TVS will continue applying the knowledge and capabilities developed through projects such as FLUTE to new clinical, research and public health challenges.
The conclusion of FLUTE is therefore not an endpoint for this work. It is another step towards a future where healthcare organisations can collaborate on AI and research while maintaining control over the data entrusted to them.
The FLUTE project has received funding from the Horizon Europe Framework Programme (HORIZON) Research and Innovation Actions under grant agreement No 101095382.
