Risk-informed framework for in silico regulatory evaluation of medical devices

This is the authors' abstract. We don't add key points for this paper.

The development of medical devices relies on empirical testing and clinical data, which cannot fully characterize safety, performance, and clinical benefit across diverse patient populations. Computational models can generate digital evidence to complement these efforts, but their integration into regulatory pathways is hindered by inconsistent requirements for model credibility and reporting. In this review, we present a risk-informed framework for the integration of in silico evidence into regulatory submissions, developed through a collaboration involving the UK Medicines and Healthcare products Regulatory Agency, industry, and academia. Using the redesign of a transcatheter aortic valve implantation device as an exemplar, we introduce a workflow to select and prioritize device-related harms for in silico assessment based on regulatory relevance, mechanistic linkage, and in silico evidence value. We define a risk-informed credibility matrix, integrating model influence and decision consequence, to establish the rigor for verification, validation, and uncertainty quantification.