Ariel Seeley served on the panel titled AI Trends in Medical Devices: Global Developments, FDA Policies, and Software as a Medical Device at the 2025 Food and Drug Law Institute (FDLI) Annual Conference. The panel provided an overview of FDA’s approach to predetermined change control plans (PCCPs) and its draft guidance on pre-market submission requirements for AI-enabled devices, a summary of current trends in devices enabled with artificial intelligence (AI) or authorized with PCCPs, as well as the burgeoning global adoption of PCCPs.
For those that missed the panel or would like a brief recap, key takeaways from the panel presentation and discussion follow.
FDA Guidance
- FDA’s guidance on the use of PCCPs for AI-enabled devices provides recommendations on establishing and modifying PCCPs, the information that should be included in a PCCP, and the types of changes best suited to a PCCP.
- FDA signifies that it is open to considering and reviewing generative AI-based modifications through a PCCP.
- For AI-enabled devices, FDA recommends that the modification protocols in the PCCP should contain a detailed description of the data management and retraining practices, performance evaluation protocols, and real-world monitoring plans.
- FDA’s draft guidance on lifecycle management and marketing submission recommendations for AI-enabled devices provides proposed recommendations on the content of marketing submissions for AI-enabled devices.
- While FDA’s draft guidance is not for implementation, it may be the best indicator of the types of information FDA is looking for in an AI-enabled device pre-market submission for some time.
- Based on the draft guidance, FDA is focused on device transparency (i.e., ensuring that users understand the outputs of AI-enabled devices), and the guidance includes an example Model Card as a way to promote consistency in how manufacturers convey information about AI-enabled devices.
- FDA indicates that the quality, diversity, and quantity of data that the AI models are trained on and tuned with are integral to bias mitigation and overall device accuracy.
- FDA remains focused on post-market performance monitoring to mitigate drift and ensure that device performance does not excessively deviate outside of its authorized specifications.
Trends in AI-Enabled Devices
- Excitement around AI and its capability to revolutionize the provision of healthcare continues to surge. Over the last five years, there has been a rapid increase in the number of AI/machine learning-enabled devices authorized by FDA.
- FDA has authorized more than 1,000 such devices. Most of these are in the radiology space, but other areas like cardiology, neurology, and anesthesiology are also starting to see increases in the number of authorized devices.
- All of these devices involve locked algorithms, meaning the device is not generating new content based on post-market device use.
- While FDA’s experience with these devices continues to grow, approaching pre-market interactions for AI-enabled devices with intention and nuance remains critical to ensure success.
Navigating FDA
- Early engagement with FDA to obtain upfront alignment on approach and development for new devices can be valuable—especially for truly novel technologies and complex PCCPs.
- PCCPs are an efficient vehicle, but they involve more upfront work than traditional pre-market submissions. Fully explore the benefits and drawbacks of PCCPs with individuals who are knowledgeable about the scope and development of PCCPs.
- Successful PCCP development and implementation may require socializing the concept across the business, including senior leadership.
- When there is friction between FDA and industry during pre-market discussions, it is often due to miscommunication.
- Clear communication around the indications for use and risks and benefits of the device, as well as precisely crafted, focused questions are essential to obtaining useful feedback in a timely manner.
- To ensure you are capitalizing on all opportunities to interact with FDA, it can help to work with an experienced second set of eyes to help ensure that your device description is appropriately focused, your questions are clearly articulated, that areas that could hinder productive FDA interactions are anticipated and addressed, and who can help ensure that the appropriate and experienced FDA staff are engaged in your meeting.
Conclusion
What makes the use of AI in medical devices exciting is that the field is constantly evolving. With user fee negotiations soon to be underway, we may see new policy initiatives or legislative proposals related to the use of AI in medical devices emerge from FDA and industry discussions. This is an area we will continue to monitor closely. Subscribe to our mailing list for the latest on programming, guidance, and legal and business developments involving the use of AI in the life sciences industry.