Regulation of Health-Related Artificial Intelligence in Medical Devices: The Canadian Story

Abstract

Artificial Intelligence (AI) may transform Canadian healthcare. The hope is that AI will enable more accurate and efficient care, thereby solving many access, quality, and safety problems. Despite this tantalizing prospect, there are risks of unsafe AI harming patients, algorithmic bias, and threats to privacy. This work begins analysis of whether applicable laws are up to the task of ensuring Canadians can benefit from effective health-related AI while minimizing AI-related risks. It focuses on Health Canada’s regulation of medical devices, a ‘first line of defence’ that decides which devices are safe, effective, and thus permitted for trade in Canadian markets. After highlighting the regulatory challenge, we provide the first detailed explanation of Canadian medical device regulations and how they apply to AI-enabled devices. We then discuss a still-developing “alternative pathway” for licencing devices with AI and the regulatory gaps left open. We conclude with recommendations that a recent emphasis on post-market surveillance should not be at the expense of robust pre-market review and that safety and efficacy review embrace bias- and privacy-related risks. Further, while post-market surveillance holds potential for ensuring the safety of adaptive machine-learning medical devices over time, much will depend on regulatory capacity and competency and investments therein.

More information

Identifiers

ORCID for Michael Da Silva: orcid.org/0000-0002-7021-9847

Catalogue record

Export record