Haematological diseases (HDs) encompass a wide range of disorders arising from quantitative or qualitative abnormalities in blood cells, lymphoid organs, and coagulation factors. Despite approximately 74% of these conditions being classified as rare, they collectively impose a significant global health burden and economic strain on healthcare systems. The scarcity of patient data and fragmentation across unconnected clinical entities often hinder effective clinical approaches, particularly for the rarest conditions.
SYNTHEMA aims to address these challenges by establishing a cross-border data hub dedicated to developing and validating innovative AI-based techniques for clinical data anonymization and synthetic data generation (SDG). This initiative seeks to expand the foundation for GDPR-compliant research in rare haematological diseases (RHDs), focusing on two representative cases: sickle-cell disease (SCD) and acute myeloid leukaemia (AML).
The project will develop a federated learning (FL) infrastructure equipped with secure multiparty computation (SMPC) and differential privacy (DP) protocols. This infrastructure will connect clinical centers that provide standardized, interoperable multimodal datasets with computing centers from academia and SMEs. The framework will facilitate the training of AI algorithms and perform SMPC-based global model aggregation in a privacy-preserving manner. The resulting synthetic data will be validated for clinical value, statistical utility, and residual privacy risks.
Additionally, SYNTHEMA will establish legal and ethical frameworks to ensure privacy by design in the collection and processing of health-related personal data, promoting ethical co-creation of algorithms. The project’s outcomes, including pipelines, standards, and data, will be openly accessible to stakeholders in healthcare, academia, and industry, contributing to existing rare disease registries.
Link to project website: https://synthema.eu/