Imagine trying to solve a puzzle where the pieces are scattered across different hospitals and written in different languages. This project builds a digital translator and a secure vault that lets researchers study heart disease data without actually moving the private records. It's like having a smart assistant that can read thousands of medical notes and find patterns without ever seeing a patient's name.
Cardiology data is fragmented, locked in unstructured formats, and restricted by strict privacy laws, making it nearly impossible to train AI models on large-scale, multi-site datasets.
A federated, privacy-preserving toolbox including a Common Data Model, a ML Data Preparation Suite, and 7 cardiology-specific language models.
If you are an AI developer dealing with fragmented medical records — this project developed a federated learning platform and synthetic data sets that allow you to train models without compromising patient privacy.
If you are a researcher dealing with unstructured clinical notes in multiple languages — this project developed 7 language models adapted to cardiology that automate data ingestion and harmonization.
If you are a hospital administrator dealing with strict European privacy regulations — this project developed a privacy-preserving toolbox that enables multi-site data use while ensuring compliance.
Based on available project data, no specific commercial pricing or licensing costs are mentioned; the project is funded by an EU contribution of EUR 7,747,905.
Yes, the tools are validated in 7 clinical sites across Europe and are designed to be generalized for other clinical areas in medicine.
The project will provide the CardioSynth open database of synthetic data for further research and AI experimentation.
The toolbox is implemented ensuring privacy-by-design and thorough compliance with European regulations and data standards.
Integration is achieved through a Common Data Model and a Data Ingestion Suite designed to harmonize diverse data formats.
The consortium is well-balanced for technology transfer, featuring 18 partners across 10 countries. With an industry ratio of 28% (including 5 industry partners and 3 SMEs), there is a strong link between the 11 academic/research entities and real-world commercial application.
Contact Universitat de Barcelona regarding the federated learning platform
Contact us to explore licensing for the CardioSynth synthetic database.
