Regulatory Readiness with Synthetic Data | Part 2: FDA and EMA Case Studies
Part 2 of a 3-part series on regulatory readiness with synthetic data
FDA Signals: From Grand Rounds to OSEL Tools
The FDA is no longer treating synthetic data as theory. Through its 2024 Grand Rounds, FDA scientists outlined how synthetic imaging can accelerate model evaluation while protecting patient privacy. This aligns with earlier recognition of data bottlenecks by the Office of Science and Engineering Laboratories (OSEL).
Case Study: VICTRE
The Virtual Imaging Clinical Trial for Regulatory Evaluation (VICTRE) replaced a large-scale breast imaging trial with fully synthetic data. It generated tens of thousands of virtual mammograms to test lesion detectability, removing the need for real-patient enrollment. The outcome: equivalent conclusions with faster timelines and lower cost (PMC12072219).
Case Study: M-SYNTH
M-SYNTH is another FDA-led initiative focused on synthetic mammography datasets. By simulating device protocols and patient variability, it provides a controlled testbed for evaluating AI algorithms. Unlike traditional datasets, M-SYNTH can be regenerated on demand, enabling reproducibility and scalability that regulators value (arXiv 2024).
EMA Perspective
The European Medicines Agency (EMA) has not yet released formal guidance, but academic literature indicates a convergence of priorities. A clinical trial review highlights synthetic data’s promise in rare diseases and pediatrics — areas where real data is nearly impossible to collect at scale. EMA scientists are watching FDA pilots closely, signaling likely adoption of similar frameworks.