Synthetic Data to Support MedTech Innovation, Medicines and Healthcare products Regulatory Agency
There has been an explosion of artificial intelligence algorithms used in healthcare. However, many are not validated, leading to concerns about bias and patient safety. The MHRA has developed a novel method to generate synthetic data as a cost-effective solution for algorithm training and validation. These high-fidelity datasets are completely artificial, but capture the complex clinical relationships found in the original data. This methodology provides regulators and MedTech innovators with a robust route for algorithm validation and regulation. It also provides the NHS and patients with assurance that algorithms used in clinical decision making are safe, effective and equitable.