Synthetic data is artificially generated data that reflects the properties and relationships found in real data. While synthetic data can play an important role in AIaMD development, its use in regulatory submissions introduces specific considerations, especially when it constitutes a substantial part of the evidence package.
The Synthetic Data for Development of AI as a Medical Device (AIaMDs) report, produced by the Medicines and Healthcare products Regulatory Agency (MHRA) and the PHG Foundation, outlines these considerations, building on and complementing existing regulatory guidance. This marks an important and exciting first step for manufacturers and notified bodies collectively navigating this evolving landscape. The report provides crucial groundwork, though more work is required to move past these preliminary ideas.
This document does not constitute official MHRA guidance; rather, it presents principles and reflections developed by an expert working group convened by the MHRA and the PHG Foundation, with support from the Regulators’ Pioneer Fund.
The report discusses key technical, ethical, regulatory, and lifecycle aspects to consider when planning to use synthetic data, and provides reflective questions to help manufacturers critically assess the quality, provenance, and suitability of their synthetic datasets throughout the AIaMD lifecycle. The core principles of fidelity, representativeness, and transparency are emphasised as essential for building regulatory confidence.
Supporting resources and relevant best practices are also included to help manufacturers align with current standards and recommendations.
Read the full Synthetic Data for Development of AI as a Medical Device (AIaMDs) report to explore these considerations in detail.
Valena Reich, Colin Mitchell, Elizabeth Redrup Hill, Puja Myles, Richard Branson, Russell Pearson and members of the Expert Working Group