Imagine needing a huge library of medical records to train a smart computer, but the books are locked in different vaults for privacy. Instead of moving the real books, this project creates perfect 'fake' copies that look and act like the real ones but contain no private info. It also lets computers learn from the real data without ever actually seeing the private details.
AI developers in healthcare cannot access the large, diverse datasets needed for training because health data is locked in silos due to strict privacy laws and distributed storage.
A system for generating synthetic medical data (EHRs and images) and a multi-party computation service for analyzing distributed data without moving it. It also includes a Data Market for the monetization of these services.
If you are a startup dealing with a lack of diverse training images for cancer detection — this project developed synthetic data pipelines that provide anonymous medical images. This allows you to train your AI without the legal risk of handling real patient files.
If you are a firm dealing with fragmented patient data across different countries — this project developed multi-party computation services. This lets you derive insights from distributed real-world data without needing to move it into a single database.
If you are a platform operator dealing with the difficulty of monetizing sensitive health records — this project developed a Data Market for sharing and monetization. It includes incentive-based systems to encourage data provision while ensuring privacy.
Based on available project data, the specific pricing is not listed, but the project aims to establish a Data Market to facilitate monetization and incentives for data provision.
Yes, the project intends to integrate the ecosystem as a cross-European health data hub within the European Health Data Space.
Based on available project data, specific licensing terms are not provided, though the project focuses on creating a market for data sharing and monetization.
The project uses synthetic microdata, anonymization, and multi-party computation to ensure privacy-compliant data utilization without compromising data subjects.
The project period runs from 2023-10-01 to 2026-12-31, suggesting the full ecosystem will be ready by the end of 2026.
The consortium is well-balanced for commercialization, consisting of 21 partners with a significant industrial presence (38% industry ratio). With 8 industry partners and 5 SMEs across 10 countries, the project has a strong bridge between academic research (8 universities, 5 research centers) and market application, particularly in the European health sector.
Contact Turku University (Finland) for technical inquiries regarding the data synthesis pipelines.
Contact us to connect with the PHASE IV AI consortium for early access to synthetic health datasets.
