How Does the FDA Address AI Algorithm Updates?

By Rasit Dinc

Introduction

Artificial intelligence (AI) and machine learning (ML) are no longer futuristic concepts in healthcare; they are increasingly integral to a new generation of medical devices. These technologies offer immense potential to improve diagnostics, personalize treatments, and enhance patient care. However, their adaptive nature, particularly the ability of algorithms to learn and evolve with new data, presents a significant challenge to traditional regulatory frameworks. The U.S. Food and Drug Administration (FDA), tasked with ensuring the safety and effectiveness of medical devices, has been proactively developing a modernized approach to address the unique complexities of AI/ML algorithm updates. This article explores the FDA's evolving strategy for regulating these dynamic technologies [1].

The Limitations of Traditional Regulation

The FDA's established regulatory pathways for medical devices, including premarket clearance (510(k)), De Novo classification, and premarket approval (PMA), were designed for static devices. These frameworks require manufacturers to seek new clearance or approval for significant modifications that could affect a device's safety or effectiveness. While this model has served well for traditional hardware and software, it is ill-suited for the iterative and adaptive nature of AI/ML-based Software as a Medical Device (SaMD). Requiring a new premarket review for every algorithm update would stifle innovation and delay patient access to improved technologies [2].

A New Paradigm: The Predetermined Change Control Plan

Recognizing the need for a new regulatory paradigm, the FDA introduced the concept of a Predetermined Change Control Plan (PCCP) in its 2019 discussion paper, "Proposed Regulatory Framework for Modifications to AI/ML-Based Software as a Medical Device" [3]. A PCCP is a plan that a manufacturer submits to the FDA as part of a premarket submission. This plan outlines the anticipated modifications to an AI/ML-based SaMD, including the specific algorithms to be updated and the methodology for implementing and validating those changes. If the FDA agrees to the PCCP, the manufacturer can make the specified updates without needing to submit a new premarket submission for each change.

This innovative approach allows for a more streamlined and efficient regulatory process, fostering the very innovation that makes AI/ML so promising. The FDA's thinking on this was further solidified in its January 2021 "Artificial Intelligence and Machine Learning Software as a Medical Device Action Plan," which outlined a multi-pronged approach to advancing the agency's oversight of AI/ML-based medical software [4].

Guiding Principles for Responsible AI Development

To support the PCCP framework and promote responsible AI/ML development, the FDA, in collaboration with international partners, has established a set of guiding principles. These include:

• Good Machine Learning Practice (GMLP): These principles, released in October 2021, provide a set of best practices for the development and lifecycle management of AI/ML-based medical devices. They emphasize the importance of robust data management, model training, and performance evaluation [5].
• Transparency: The FDA has stressed the importance of transparency in AI/ML-based devices. In June 2024, the agency released guiding principles on transparency, which encourage manufacturers to provide clear information to users about the device's intended use, its performance, and the data used to train the algorithm [6].
• Marketing Submission Recommendations for a PCCP: In December 2024, the FDA issued final guidance on the marketing submission recommendations for a PCCP. This guidance provides manufacturers with a clear roadmap for what to include in their PCCP submissions [7].

The Road Ahead: A Lifecycle Approach

The FDA's approach to AI/ML regulation continues to evolve. In January 2025, the agency published a draft guidance on the lifecycle management of AI-enabled device software functions. This guidance emphasizes a holistic, risk-based approach that spans the entire product lifecycle, from initial design and development to post-market surveillance. It also highlights the importance of a collaborative approach, with the FDA's Center for Biologics Evaluation and Research (CBER), Center for Drug Evaluation and Research (CDER), Center for Devices and Radiological Health (CDRH), and Office of Combination Products (OCP) working together to ensure a consistent regulatory approach across all medical products [8].

Conclusion

The FDA's approach to regulating AI algorithm updates is a testament to its commitment to fostering innovation while safeguarding public health. By moving away from a one-size-fits-all regulatory model and embracing a more flexible, risk-based framework centered on the Predetermined Change Control Plan, the FDA is paving the way for the safe and effective integration of AI/ML into the healthcare landscape. As these technologies continue to advance, the FDA's adaptive and collaborative approach will be crucial for ensuring that patients can benefit from the full potential of AI-powered medical devices.

References

[1] U.S. Food and Drug Administration. (2025, March 25). Artificial Intelligence in Software as a Medical Device. FDA. https://www.fda.gov/medical-devices/software-medical-device-samd/artificial-intelligence-software-medical-device

[2] Wu, K., Wu, E., Rodolfa, K., Ho, D. E., & Zou, J. (2024). Regulating ai adaptation: An analysis of ai medical device updates. arXiv preprint arXiv:2407.16900.

[3] U.S. Food and Drug Administration. (2019). Proposed Regulatory Framework for Modifications to Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD). FDA. https://www.fda.gov/files/medical%20devices/published/US-FDA-Artificial-Intelligence-and-Machine-Learning-Discussion-Paper.pdf

[4] U.S. Food and Drug Administration. (2021). Artificial Intelligence and Machine Learning Software as a Medical Device Action Plan. FDA. https://www.fda.gov/media/145022/download

[5] U.S. Food and Drug Administration. (2021). Good Machine Learning Practice for Medical Device Development: Guiding Principles. FDA. https://www.fda.gov/medical-devices/software-medical-device-samd/good-machine-learning-practice-medical-device-development-guiding-principles

[6] U.S. Food and Drug Administration. (2024). Transparency for Machine Learning-Enabled Medical Devices: Guiding Principles. FDA. https://www.fda.gov/medical-devices/software-medical-device-samd/transparency-machine-learning-enabled-medical-devices-guiding-principles

[7] U.S. Food and Drug Administration. (2024). Marketing Submission Recommendations for a Predetermined Change Control Plan for Artificial Intelligence-Enabled Device Software Functions. FDA. https://www.fda.gov/medical-devices/software-medical-device-samd/marketing-submission-recommendations-predetermined-change-control-plan-artificial-intelligence

[8] U.S. Food and Drug Administration. (2025). Draft Guidance: Artificial Intelligence-Enabled Device Software Functions: Lifecycle Management and Marketing Submission Recommendations. FDA. https://www.fda.gov/media/184856/download