FDA RELEASES DRAFT AI GUIDANCE FOR MEDICAL DEVICES
MARKETING SUBMISSION RECOMMENDATIONS FOR PREDETERMINED CHANGE CONTROL PLANS FOR AI- AND ML-ENABLED DEVICE SOFTWARE FUNCTIONS
On March 30, 2023, the U.S. Food and Drug Administration (FDA) issued the draft machine learning and AI guidance for medical devices: “Marketing Submission Recommendations for a Predetermined Change Control Plan for Artificial Intelligence/Machine Learning (AI/ML)-Enabled Device Software Functions.” Outlined within the document is the use of Predetermined Change Control Plans (PCCPs) for AI/ML software and incorporates comments and feedback from the April 2019 discussion paper, “Proposed Regulatory Framework for Modifications to Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD).”
The development and release of this draft guidance represents the next step in the FDA’s work to develop a regulatory framework for AI/ML-based software as described in the agency’s January 2021 document, “FDA’s Artificial Intelligence/Machine Learning (AI/ML) Software as a Medical Device Action Plan.” The issuance of this draft guidance demonstrates the FDA’s broader commitment to developing innovative approaches to the regulation of device software functions.
The aim of the AI guidance for medical devices is to address the challenge of adaptive AI/ML, including the current requirement that manufacturers must resubmit to the FDA for device modifications when software undergoes anticipated algorithm updates. This guidance will allow manufacturers and the FDA to agree on anticipated post-approval product updates as part of pre-market review (PCCPs are currently employed on a limited basis by FDA).
Presently, the FDA has legal authority to employ PCCPs for De Novos, however the Food and Drug Administration Safety and Landmark Advancements Act of 2022 would give the FDA explicit authority to approve PCCPs as part of the 510(k) and Pre-Market Approval (PMA) process.
As it stands today, most AI/ML devices are marketed with locked algorithms. In contrast with other AI technologies, the algorithms learn without training data. Through reason, deduction and inference, device-specific AI may evolve or adapt, refine its performance, or calibrate itself to a specific patient or health care setting’s characteristics during use. AI that learns and changes itself during clinical use requires manufacturers to determine the limits for such change to occur safely while assuring performance for the intended purpose.
Draft Guidance Call Outs
The FDA’s draft guidance includes proposed recommendations for the content to include within PCCPs as part of a marketing submission for AI/ML-enabled devices.
- A PCCP is to outline anticipated device changes and how they will be assessed and implemented in accordance with the PCCP (among other items).
- New definitions are introduced, such as machine learning-enabled device software functions (ML-DSFs).
- The scope of this guidance focuses on ML-DSFs, including information on:
– What modifications to the ML model are implemented automatically by software; and
– For ML-DSFs, what modifications to the ML model are implemented manually (i.e., involving steps that require human input, action, review, and/or decision-making, and therefore, not implemented automatically).
- The FDA has proposed the PCCP concept for not only AI/ML-enabled SaMD, but for all AI/ML-enabled device software functions. This includes software functions that are part of/or control hardware medical devices.
Other changes outside of PCCP that contain minor modifications (not requiring a new submission) are outside the scope of this guidance. The FDA still points to the review of:
- Deciding When to Submit a 510(k) for a Software Change to an Existing Device | FDA
- Deciding When to Submit a 510(k) for a Change to an Existing Device | FDA
- Modifications to Devices Subject to Premarket Approval (PMA) – The PMA Supplement Decision-Making Process | FDA
The FDA also considers the PCCP to be part of the technological characteristics of a device and explains that where a predicate device was authorized with a PCCP, the subject device must be compared to the version of the predicate device—cleared or approved prior to changes made under the PCCP. Combination products (such as drug-device and biologic-device combination products) when the device constituent part is, or includes, an ML-DSF are also in scope of this draft guidance.
Pre-market authorization for an ML-DSF with a PCCP must be established through the 510(k) pathway, De Novo pathway or PMA pathway. A PCCP must be reviewed and established as part of a marketing authorization for a device prior to a manufacturer making modifications under that PCCP.
The FDA encourages manufacturers to leverage the Q-Submission process for obtaining FDA feedback on proposed PCCPs prior to marketing application submissions. The agency also provides details on how and where this information should reside or be discussed in the marketing application (Cover Letter, TOC, Device Description and labeling, to name a few).
Finally, the draft guidance includes performance considerations with respect to race, ethnicity, disease severity, gender, age and geographical consideration. This is part of the ongoing development, validation, implementation and monitoring of AI/ML-enabled devices. Notably, the draft guidance proposes to place a significant and increased emphasis on the importance of clearly communicating valuable information in the labeling.
The draft guidance is open for comment for 90 days, until July 3, 2023. The FDA welcomes continued feedback through public docket. Commenters should visit www.Regulations.gov and search for docket number FDA-2022-D-2628; comments may be submitted by electronic or written comment before July 3 to ensure that all feedback is considered before the FDA releases the final guidance.
On April 13, 2023, the FDA will host a webinar for industry, healthcare providers and others interested in learning more about the draft guidance: Webinar – Marketing Submission Recommendations for a Predetermined Change Control Plan for Artificial Intelligence / Machine Learning-Enabled Device Software Functions Draft Guidance – 04/13/2023 | FDA
How Can NAMSA Help?
NAMSA experts possess in-depth experience working with medical device manufacturers who make a wide variety of patient monitoring, disease management, PACS imaging and other software-containing medical devices whose value and effectiveness are enhanced through mHealth technology. We offer regulatory and quality consulting services for medical devices, software as a medical device (SaMD), IVD and digital health sectors.
NAMSA’s regulatory experts regularly provide consulting services to organizations who design, produce, manufacturer, supply or deploy the following:
- Software as a Medical Device (SaMD)
- Mobile Medical Apps
- Medical devices of all types with a particular focus on “active” devices and IVDs with or without software components or accessories
- Clinical Decision Support and health analytical software
- Software as a Service (SaaS) within the healthcare sector
- Artificial Intelligence (AI), deep learning, machine learning and big data algorithms
Learn about NAMSA’s full suite of Regulatory and Quality services and solutions.
Or, contact our team if you are interested in meeting with one of our regulatory experts.
Monica R. Montanez
Monica R. Montanez, MS, RAC, CQA currently serves as NAMSA's Principal Regulatory Consultant. Monica has over twenty years’ experience in the medical device industry in Regulatory Affairs and Quality Assurance. Her primary focus is navigating the regulatory pathways for electro-mechanical and software driven medical devices worldwide. She has received clearance of many 510(k)s and approval of new indications for PMA device(s) of which 90% involved software. More recently, she has broadened her regulatory experience in the area of digital health that includes: Software as Medical Device (SaMD), Mobile Medical Apps (MMA), Digital Therapeutics(DTx), Artificial Intelligence (AI), Machine Learning (ML), Cybersecurity, Usability, and Risk Management. While in industry, she assisted in the development of FDA 510(k) guidance and FDA Software guidance directly with FDA. Monica holds a Masters of Science (MS) degree in Regulatory Science (RS) from the University of Southern California (USC) School of Pharmacy. Currently. she holds Regulatory Affairs Certification (RAC) from the Regulatory Affairs Professionals Society (RAPS) and Certified Quality Auditor (CQA) from the American Society for Quality (ASQ).