Strategies for formulating a structured modification management scheme

Strategies for formulating a structured modification management scheme

The Food and Drug Administration (FDA) has finalized its guidance on Pre-Determined Change Control Plans (PCCPs) for medical devices incorporating artificial intelligence (AI), expanding the scope of PCCPs to cover all AI-driven technologies. This move, announced in December 2024, represents a significant step forward in the regulation of AI medical devices.

Under the FDA's framework, AI device manufacturers are now required to submit a proactive plan that outlines the types of algorithm changes expected post-market, methods for implementing, testing, and validating those changes, and risk mitigation strategies to ensure continued safety and effectiveness. Once the PCCP is approved, manufacturers can iterate and improve their AI models faster while maintaining continuous regulatory oversight and ensuring patient safety.

The implications of PCCPs for AI medical devices are far-reaching. They promise faster innovation cycles, enabling manufacturers to update AI systems within pre-approved boundaries without the need for repeated full FDA submissions. This expedites improvements and adaptation to new clinical data or technology advances.

PCCPs also enhance patient safety and transparency, requiring performance monitoring and risk management, supporting the safe evolution of AI devices over time. Furthermore, the FDA's expansion of PCCPs to cover all AI technologies, including foundation models such as large language models, demonstrates its readiness to accommodate diverse AI architectures in healthcare.

The FDA's emphasis on Total Product Life Cycle (TPLC) and the use of real-world clinical data complements PCCPs, facilitating ongoing device validation post-market. This shift towards continuous oversight models reflects the FDA's commitment to balancing innovation and safety for adaptive AI medical devices, positioning the regulatory system to support rapid but controlled AI development in healthcare.

Beacon Biosignals, a company that has successfully navigated the PCCP process, emphasizes the importance of a validation plan showing safety and effectiveness. This may include updating device labeling to reflect changes in algorithm speed or sensitivity/specificity. Companies must submit a modification protocol to the FDA and follow the same procedure every time they make changes with PCCPs.

It is important to note that changes made with PCCPs must be necessary to maintain or improve a product's safety and effectiveness and cannot change the intended use of a device. The FDA's authority to clear and approve PCCPs comes from the 2022 omnibus spending bill.

However, the nature of AI makes PCCPs a topic of discussion for most, if not all, AI- or ML-enabled device functions. The FDA's draft guidance aims to boost transparency and stem AI device bias. Lack of detail is a concern for the FDA, and early conversations with the agency can mitigate critiques.

Some manufacturers have not yet adopted PCCPs due to discomfort with the framework or lack of planning. Hospital leaders and manufacturers should closely engage with PCCPs to optimize AI device deployment and maintenance. Companies should develop a plan for notifying customers about changes after a product is on the market.

PCCPs are expected to remain despite regulatory uncertainty surrounding AI in medical devices under the current administration. They could facilitate more automated software updates for medical devices, making the process more efficient and streamlined.

For companies planning to use PCCPs for medical devices, it is crucial to have a detailed plan for updates. Developers must plan for the life cycle of data, how changes will be managed, performed, and evaluated, and document these plans. The FDA's Q-submission program offers an opportunity for feedback before submitting a PCCP.

In short, PCCPs are now established, FDA-endorsed tools for managing planned algorithm changes in AI-enabled medical devices throughout their lifecycle, enabling safer, faster, and more transparent innovation. By engaging with PCCPs, hospital leaders and manufacturers can optimize AI device deployment and maintenance, ensuring the continued safety and effectiveness of AI medical devices.

1. The Food and Drug Administration (FDA) has expanded the scope of Pre-Determined Change Control Plans (PCCPs) to cover all AI-driven medical technologies.
2. Under the FDA's framework, AI device manufacturers are now required to submit a proactive plan that outlines expected post-market algorithm changes.
3. PCCPs promise faster innovation cycles for AI medical devices, enabling updates within pre-approved boundaries without the need for repeated full FDA submissions.
4. The FDA's emphasis on Total Product Life Cycle (TPLC) and the use of real-world clinical data complements PCCPs, facilitating ongoing device validation post-market.
5. Beacon Biosignals, a company that has successfully navigated the PCCP process, emphasizes the importance of a validation plan showing safety and effectiveness.
6. Changes made with PCCPs must be necessary to maintain or improve a product's safety and effectiveness and cannot change the intended use of a device.
7. For companies planning to use PCCPs for medical devices, it is crucial to have a detailed plan for updates and to consider the life cycle of data, how changes will be managed, and document these plans.
8. PCCPs are expected to remain despite regulatory uncertainty surrounding AI in medical devices under the current administration, potentially making the process more efficient and streamlined for medical device updates.

Read also: