At the ICT&health World Conference 2026 in Maastricht, SYNTHIA hosted a high-level panel discussion exploring the central question: Can synthetic data accelerate AI-driven innovation in healthcare while safeguarding privacy, trust, and scientific rigor?
The answers from our panel members were clear: Europe’s scientific community is hungry for data, and synthetic data may become a crucial bridge toward unlocking its full potential. Watch the SYNTHIA impression video and read the individual contributions.
Data as Europe’s strategic asset
Artificial intelligence in healthcare depends on access to high-quality, well-structured data. Yet across Europe, health data remains fragmented, sensitive, and difficult to share. Strict GDPR safeguards, heterogeneous standards, and interoperability gaps between hospitals create structural barriers for researchers and innovators.
Szymon Bielecki, Head of Sector for Research and Innovation at European Commission, positioned synthetic data within Europe’s broader digital and AI strategy. While the European Health Data Space (EHDS) aims to improve secondary use and interoperability, real-world data access remains complex. Until harmonized access mechanisms are fully operational, synthetic data may serve as a bridge. It can create realistic testing environments, support AI model development, and enable medicine innovation in privacy-preserving settings. As Bielecki emphasized, artificial intelligence will never work without good quality data. Synthetic data, however, is not a replacement for real-world data infrastructure. It is an enabling technology, one that must operate within clear governance and trust frameworks.
“Data is the oil of innovation and the foundation of artificial intelligence. But in healthcare, accessing high-quality data is particularly difficult due to sensitivity, regulatory requirements, and lack of interoperability. In this context, synthetic data can play a crucial role by creating realistic testing and training environments where real data is not yet accessible.”
Validity, utility, privacy: the three pillars
From a research perspective, synthetic data is not valuable by default. It must meet strict scientific criteria. SYNTHIA Project Coordinator Leonor Cerdá Alberich, La Fe Health Research Institute, highlighted three essential pillars: Validity (Synthetic datasets must capture complex, nonlinear relationships found in real patient data), Utility (AI models trained on synthetic data must perform reliably when applied to real-world data) and Privacy (Synthetic data must withstand adversarial testing and re-identification attacks). The distinction between statistical similarity and clinical usability emerged as a recurring theme. Synthetic data must be clinically meaningful, not just mathematically convincing. This principle lies at the core of SYNTHIA’s work.
“One of the main uses of synthetic data in medical imaging has been to cover missing information in our data warehouses. Hospitals generate huge volumes of imaging data, MR, CT, mammography, and ultrasound, but much of it is not fully leveraged by AI. Synthetic data will allow us to fill the gaps in these large hospital silos, which are often disconnected, and generate datasets that are simply not there.”
Synthetic data as a clinical accelerator
Several speakers illustrated how synthetic data can address real clinical challenges. SYNTHIA Partner Hongxu Yang, GE Healthcare, described AI systems as fundamentally “data hungry.” Collecting and annotating large-scale imaging datasets is costly and complex. Synthetic data can enhance robustness, reduce development time, and improve model stability, particularly when guided by clinical expertise.
“AI development is fundamentally data hungry, you can never have sufficient data to train a perfect model. In medical imaging especially, collecting, cleaning, and annotating data is extremely complex and expensive. Synthetic data gives us a way to build more stable solutions, particularly when we integrate clinical knowledge into the development process.”
Synthetic Data: A technological solution to clinical challenges
SYNTHIA Partner Saverio D’Amico, TRAIN / Humanitas Research Hospital, described synthetic data as a “technological solution to clinical problems”, a form of “plastic data” that can generate condition-specific cohorts once trained on real-world information. This capability opens new possibilities in rare diseases and personalized medicine. For example, in highly sensitive pediatric oncology scenarios, traditional control arms can be ethically challenging. Synthetic control arms may help strengthen statistical power where evidence is scarce. As D’Amico concluded, synthetic data is a “multi-knife technology”, versatile, powerful, and requiring careful handling.
“Synthetic data is a technological solution to clinical problems. Technology should enhance the quality of care and, ultimately, the quality of life, and synthetic data is one of the tools that can help us achieve that. In areas like rare diseases and personalized medicine, this technology can be particularly effective.”
A double-edged sword: ethical and governance challenges
SYNTHIA Partner Davide Cirillo, Barcelona Supercomputing Center, described synthetic data as a “double-edged sword.” While ethically attractive due to privacy preservation, it can also amplify biases present in the original data. Apparent neutrality does not guarantee fairness. Accountability raises further questions. If synthetic data contributes to flawed decisions, who bears responsibility, data collectors, model developers, or deploying institutions? Regulatory clarity under frameworks such as the EU AI Act remains an open challenge. Synthetic data, the panel agreed, must be treated as a socio-technical asset. That means impact assessments, responsible design, transparency, and patient awareness. Trust depends on openness.
“Synthetic data is truly a double-edged sword. It is ethically attractive because it can enhance privacy, improve data access, and even help mitigate bias. But if the original data is biased or poorly assessed, synthetic data will simply propagate those flaws. It is not neutral and we must acknowledge that responsibility and governance cannot be an afterthought.”
Breaking the Data Barrier
Sara Okhuijsen, OASYS NOW, addressed another dimension: innovation barriers for startups. Early-stage companies face a “cold start problem”, they need data to validate their solutions, yet access to data requires prior validation and trust. Synthetic data can help break this cycle while respecting privacy constraints.
“Innovation can really be sped up with synthetic data. It can help startups move faster, validate earlier, and build trust before accessing sensitive real-world datasets.”
How SYNTHIA is advancing trustworthy synthetic data in Europe
At SYNTHIA, we are working to ensure that synthetic data becomes a scientifically robust and clinically meaningful component of Europe’s digital health ecosystem. Launched in September 2024 as the first synthetic data project under the Innovative Health Initiative (IHI), we are building the infrastructure, validation frameworks, and real-world use cases needed to generate synthetic data that is privacy-preserving, reliable, and fit for clinical research.
Through our federated network of hospitals across Europe, we develop and validate synthetic data across multiple modalities, including imaging, genomics, clinical records, and laboratory data, in six disease areas: lung cancer, breast cancer, multiple myeloma, diffuse large B-cell lymphoma, Alzheimer’s disease, and type 2 diabetes.
The panel discussion at ICT&health reinforced what we see across our work: synthetic data has the potential to unlock Europe’s fragmented health data landscape and accelerate AI-driven innovation. Its long-term impact, however, will depend on maintaining scientific rigor, ethical safeguards, regulatory clarity, and public trust, principles that guide our project every day.
