Drafting a Predetermined Change Control Plan for AI/ML Medical Devices: A Practical Guide

Why Your AI/ML Device Needs a Predetermined Change Control Plan Before You Submit

If your medical device incorporates an artificial intelligence or machine learning component, you already know that the model does not stand still. It learns, drifts, and improves. The regulatory challenge is that the traditional FDA paradigm -- submit a change, wait for clearance -- was designed for static hardware, not adaptive algorithms. That mismatch creates real business risk: every meaningful model update could theoretically require a new 510(k), PMA supplement, or De Novo request.

The Predetermined Change Control Plan (PCCP) is FDA's formal answer to that problem. Used correctly, it lets you define, in advance, the types of changes you expect to make and the controls you will apply -- so those changes can be implemented without a new premarket submission. This guide gives you a practical framework for drafting one that will survive FDA scrutiny.

The Regulatory Foundation You Must Understand

The statutory authority for PCCPs was codified in Section 3308 of the Food and Drug Omnibus Reform Act of 2022 (FDORA), which amended section 515C of the FD&C Act. FDA formalized its expectations in the guidance document 'Predetermined Change Control Plans for Artificial Intelligence/Machine Learning-Enabled Devices: Guiding Principles' (2023) and the more detailed 'Marketing Submission Recommendations for a Predetermined Change Control Plan for Artificial Intelligence and Machine Learning-Enabled Device Software Functions', finalized in December 2024.

Under 21 CFR Part 820 and the quality system framework, any software change must still be evaluated under your design change procedures. A PCCP does not eliminate your QMS obligations -- it pre-authorizes specific, bounded changes at the regulatory submission level, provided your internal change controls are executed as described.

The Three Core Components FDA Expects

1. Description of Planned Modifications

This section must clearly articulate what types of changes the PCCP covers. FDA distinguishes between changes to the SaMD Pre-Specifications (SPS) -- the 'what' -- and the Algorithm Change Protocol (ACP) -- the 'how.' In the current framework, these map roughly to the Modification Protocol and the Impact Assessment within the PCCP structure.

Be specific. Saying 'the model may be retrained on new data' is not sufficient. You need to define the target use population, the input data types, acceptable performance boundaries, and whether the intended use itself could shift. FDA will push back on vague scope language because vague scope creates undefined risk.

2. Modification Protocol

The Modification Protocol is your technical and clinical roadmap for implementing the described changes. A well-drafted protocol addresses:

• Data governance: sources, curation standards, labeling procedures, and bias controls for training and validation datasets
• Performance metrics and pre-specified acceptance thresholds tied to your device's intended use and patient population
• Re-validation methodology, including holdout sets, prospective data, and real-world performance monitoring triggers
• Transparency mechanisms -- how users and clinicians will be notified of meaningful model updates

Reference your verification and validation procedures explicitly. FDA reviewers will look for traceability between the protocol and your 21 CFR 820.30 design controls documentation.

3. Impact Assessment

This is where many first-time PCCP drafters underinvest. The Impact Assessment must demonstrate that the planned modifications, even at their outer boundary, do not introduce new risks that would require a new submission. That means a credible risk analysis -- grounded in ISO 14971:2019 principles -- that evaluates the maximum foreseeable change within each modification category.

If your device cleared as a 510(k) with a substantial equivalence predicate, you must show that the changes remain within the technological characteristics and intended use envelope of that clearance. For PMA devices, the bar is higher: you must affirmatively show that the modifications do not diminish reasonable assurance of safety and effectiveness.

Practical Drafting Pitfalls to Avoid

• Overly broad modification scope: Attempting to pre-authorize too wide a range of changes signals to FDA that you have not adequately bounded your risk. Tighter scope with well-justified controls is more credible than expansive scope with weak rationale.
• Disconnecting the PCCP from your QMS: The PCCP is not a standalone document. It must integrate with your design change procedures, CAPA system, and post-market surveillance plan under 21 CFR 820.
• Ignoring the transparency requirement: FDA expects manufacturers to communicate meaningful model changes to users. Failure to address this in your PCCP is a red flag during review.
• Treating the PCCP as a one-time exercise: The plan itself can be updated via a subsequent premarket submission. Build in a process for PCCP amendments as your device evolves.

Submission Strategy: Where Does the PCCP Live?

A PCCP can be included in an initial 510(k), De Novo, or PMA submission, or submitted as a supplement or amendment to an existing cleared or approved device. FDA recommends using a Q-Submission (pre-submission meeting) to align on scope before you invest significant resources in drafting -- particularly for novel AI architectures or adaptive learning systems that update at the point of care.

The Bottom Line

A well-executed PCCP is a competitive asset, not just a compliance exercise. It shortens your post-market iteration cycle, reduces regulatory unpredictability, and signals to FDA -- and to investors -- that your organization understands how to govern AI responsibly. But it requires precise drafting, deep integration with your quality system, and a realistic risk analysis that reflects your actual development pipeline.

At ADB Consulting & CRO Inc., Andre Butler and the team work directly with AI/ML device manufacturers to develop PCCPs that are technically rigorous and strategically aligned with your go-to-market timeline. Whether you are preparing an initial submission or retrofitting a PCCP onto an existing cleared device, we can help you get it right the first time.

Ready to build a PCCP that stands up to FDA review? Book a free discovery call with our team at adbccro.com and let's map out your regulatory strategy today.