FLAME: A Platform for Privacy-Compliant Federated Learning and Analyses in Medicine 

FLAME is an open-source platform developed within the German PrivateAIM project to enable privacy-compliant, auditable, and institution-wide analyses of sensitive medical data. This presentation offers an overview of the FLAME architecture, including the hub-and-node design, orchestration mechanisms, and the integrated Policy Framework that manages permissions, conditions, and data-access rules. Using a practical example from the initial test deployment on testing infrastructure, the presentation shows how FLAME allows federated analytics without requiring patient-level data to leave the local institution. 

• Understand the core FLAME architecture, including the hub-and-node model, message brokering, storage, orchestration, and modular service components.

• Explain the FLAME Policy Framework, covering how permissions, policies, and the PDP/PIP/PE workflow work together to enforce privacy and regulatory requirements.

• Describe how federated analyses are conducted in practice, including communication flows, analysis execution, logging, and auditing.

• Identify the key benefits of FLAME for research and clinical settings, including privacy protection, interoperability, scalability, and compliance with the GDPR, MII, and institutional governance.