Scaling Federated AI for the NHS in Secure Research Environments

We're excited to announce that Flower is now fully deployable in Secure Research Environments (SREs), also known as Trusted Research Environments or Secure Data Environments. A network of SREs have been established by the NHS for scalable and safe data sharing. This nation-wide UK resource offers computational resources to researchers and practitioners that they can trust to use in order to analyze sensitive medical-related data. Now that Flower is compatible with this SRE network, a major roadblock to building large-scale NHS-based federated AI solutions has been removed.

In this blogpost, we highlight a deployment of Flower in the Secure Research Computing Platform (SRCP) at the University of Cambridge, which is one of the UK's leading NHS-compliant trusted research environments. This deployment brings the power of federated AI to high-security research environments compliant with the NHS's data security standards. Flower SuperNodes running in SRCPs enable the NHS to securely participate in an international federation of 100+ hospitals via the BloodCounts! project, collaborating in the training of AI models for predicting haematological diseases, including iron deficiency and leukemia.

Flower SuperNodes Now Available in SRCP Environments

While Flower SuperNodes are typically deployed using Flower Labs' official container images on Docker Hub, environments like SRCP require container solutions that align with their strict security and compatibility policies. To meet these requirements, Flower is now fully compatible with Apptainer (formerly Singularity) — the open-source container platform widely adopted by high-performance computing centers and research institutions worldwide.

Deploying a SuperNode in SRCP is straightforward, secure, and IT-approved:

• Step 1: Coordinate with SRCP IT Support

  Engage with the SRCP IT team to align on your project goals, network requirements, and expected workflow. Their experts will guide the process, manage vulnerability assessments, and handle necessary security approvals.

• Step 2: Convert Flower container images to Apptainer

  Pull the official Flower SuperNode container image in a secure environment and convert it into a Singularity Image Format (.sif). Once converted, the image can be safely transferred to the SRCP environment:

  Copy

  apptainer pull flower-supernode.sif docker://flwr/supernode:latest

• Step 3: Connect your SuperNode to a Flower federation

  Launch your SuperNode using standard Apptainer commands. The command structure mirrors the Docker version, ensuring a smooth transition for Docker users. Furthermore, since Flower SuperNodes operate in an egress-only network, which only allows outbound connections, this setup is ideal for SREs that require this network configuration because it eliminates the need to expose ports and makes clients effectively invisible to external systems.

  Copy

  apptainer exec flower-supernode.sif # + other SuperNode configs

Collaborative and Privacy-Preserving Medical AI

Deploying Flower within SREs like SRCP establishes a secure and compliant foundation for federated learning in healthcare, enabling the NHS and global health systems to collaborate on medical AI research without transferring sensitive patient data. This milestone brings federated learning into real-world clinical diagnostics, allowing hospitals to contribute to large-scale AI development while retaining full control of their data and overcoming technical and regulatory barriers. Led by Fan Zhang and the BloodCounts! team, the initiative sets a new benchmark for secure, privacy-preserving, and ethically responsible medical AI collaboration.

About BloodCounts!

BloodCounts! is a global consortium of leading healthcare and research institutions working together to advance AI-driven analysis of Full Blood Count (FBC) data for early disease detection. By exploring Federated Learning and machine learning, the consortium aims to drive innovation and improve global healthcare through privacy-preserving, data-driven research. Its members include clinicians and academic partners from world-renowned institutions such as: Cambridge University Hospitals NHS Trust (UK), NHS Blood and Transplant (UK), University College London Hospitals NHS Foundation Trust (UK), Barts Health NHS Trust (UK), Amsterdam University Medical Centres (The Netherlands), Maastricht University Medical Centre (The Netherlands), Zuyderland MC (The Netherlands), Health Services Authority Singapore (Singapore), Apollo Hospitals (India), The West African Centre for Cell Biology of Infectious Pathogens (Ghana), MRC Unit The Gambia (Gambia), University of Cambridge (UK), University College London (UK), KU Leuven (Belgium), and VU Amsterdam (The Netherlands).

Related Articles

• Blogpost: Announcing BloodCounts! and Flower Partnership
• Use cases: Federated Learning in Healthcare