The FDA has recently released a new draft guidance on “Marketing Submission Recommendations for a Predetermined Change Control Plan for Artificial Intelligence/Machine Learning (AI/ML)-Enabled Device Software Functions”. This draft guidance proposes a novel and flexible regulatory approach for AI/ML-based medical devices, inviting public comments until July 3, 2023.
A significant tool that can help in comprehending this draft guidance is the FDA’s recorded webinar, available on their YouTube channel. The webinar meticulously dissects the proposed recommendations, providing viewers with a comprehensive understanding of the draft guidance.
What the Webinar Reveals
The FDA webinar sheds light on some of the critical aspects of the draft guidance. It offers a detailed explanation of the Predetermined Change Control Plan (PCCP), a strategic roadmap outlining planned modifications to AI/ML-enabled device software functions, the methodology for implementing and validating these changes, and an impact assessment of these modifications. The webinar also addresses the challenges of the traditional regulatory approach to hardware-based medical devices and the need for a more flexible regulatory framework for software devices.
A New Regulatory Approach
The draft guidance marks a significant shift in the FDA’s traditional regulatory approach. The rapid, iterative design and development of software device functions, including Software as a Medical Device, necessitate a different regulatory framework. This new approach acknowledges the potential of AI/ML technologies to optimize performance over time and adapt based on real-world experience.
The Predetermined Change Control Plan (PCCP) Unpacked
At the heart of the FDA’s draft guidance lies the concept of the Predetermined Change Control Plan (PCCP). This plan is devised to map out prospective modifications in AI/ML-enabled device software functions, in turn, illustrating the process of realizing and verifying these changes, as well as assessing their impacts.
The PCCP is designed to put manufacturers on the front foot, allowing them to pre-empt changes to AI/ML algorithms and execute post-market modifications. This framework fosters an ethos of continual refinement and adjustment based on actual experiences, a fundamental advantage of AI/ML technologies.
For compliance with the PCCP, manufacturers are required to furnish:
- A comprehensive depiction of the specific modifications planned for the device.
- An all-encompassing methodology to execute, authenticate, and introduce these modifications, all the while ensuring the enduring safety and efficacy of the devices.
- An assessment of impacts, evaluating the advantages, risks, and requisite mitigations connected with the planned modifications.
The PCCP’s relevance is not confined to AI/ML-enabled Software as a Medical Device. It encompasses all AI/ML-enabled device software functions, extending to software functions that are integral to or manage hardware medical devices. This plan paves the way for manufacturers to maintain the currency of their medical devices with the latest AI/ML advancements, without the need for a new clearance or authorization for each adjustment, provided the changes are consistent with the predetermined plan.
This approach is designed to enable AI/ML-enabled medical devices to continuously learn and adapt, enhancing their performance while maintaining safety and efficacy. It aims to foster innovation in healthcare, while balancing the protection of public health.
In conclusion, the FDA’s webinar serves as a valuable resource for anyone involved in or interested in the AI/ML healthcare technology space, providing a detailed overview of the draft guidance and its implications.