JMIR Publications

ABSTRACT

The capability of artificial intelligence (AI) and machine learning (ML) to enhance performance through continuous learning presents significant advantages. However, current regulatory frameworks are based on traditional and slowly updated devices and require cumbersome documentation revisions and revalidation for updates. This poses challenges for manufacturers, leading to poor market uptake and hindering the potential benefits of AI/ML technologies. Recognizing these challenges, regulatory agencies have introduced guidelines to facilitate the approval of continuously evolving AI/ML technologies. This article examines predetermined change control plans (PCCPs) proposed by the U.S. Food and Drug Administration (FDA). PCCPs provide a structured regulatory approach that allows pre-approved, controlled AI/ML model updates while maintaining compliance and safety. PCCPs oversight tool developers’ quality management systems, define algorithm change protocols, and foster innovation while ensuring safety and regulatory compliance.