One of the fundamental tensions in AI medical device regulation is that machine learning models are designed to change. Retraining on new data, adapting to new patient populations, improving performance as more clinical evidence accumulates — these are features, not bugs. But FDA's traditional 510(k) framework treats any significant change to a cleared device as requiring a new submission.

The Predetermined Change Control Plan, or PCCP, is FDA's answer to this tension.

PCCP DECISION FLOW: IS A NEW 510(K) REQUIRED? Algorithm change planned Change within PCCP scope? YES No new 510(k) Validate per PCCP protocol and deploy NO New 510(k) required Or expand PCCP via supplement A PCCP must be submitted with the original 510(k) clearance. It cannot be added retroactively without a new submission.
The PCCP decision flow for a planned algorithm change. Changes within the approved PCCP scope require internal validation only — no FDA review.

The regulatory basis for PCCPs

FDA formalized the PCCP framework in its 2023 guidance document "Marketing Submission Recommendations for a Predetermined Change Control Plan for Artificial Intelligence-Enabled Device Software Functions." The guidance established that PCCPs could be submitted as part of a 510(k) or De Novo application, and defined what a complete PCCP submission must include.

A PCCP, reviewed and approved by FDA as part of the initial clearance, constitutes advance FDA agreement that the specified modifications do not require a new submission. It's a pre-negotiated change approval, not an exemption from oversight.

What a PCCP contains

A complete PCCP submission has three required components: the Description of Modifications (exactly what types of algorithm changes are anticipated), the Modification Protocol (how changes will be implemented and validated), and the Impact Assessment (analysis of whether planned modifications could introduce new safety concerns).

What types of changes PCCPs typically cover

Based on approved PCCPs, modifications FDA has been willing to pre-approve include: retraining on expanded datasets, performance threshold updates, new output formats, and adaptation to new imaging hardware. Changes that are consistently outside PCCP scope include changes to intended use, fundamental algorithm architecture changes, and changes that introduce new safety risks.

Patterns in approved PCCPs

Approved PCCPs tend to be specific and bounded. FDA has not approved open-ended PCCPs that give manufacturers broad discretion over future changes. Validation requirements are substantial — the PCCP doesn't eliminate the validation burden, only the FDA review burden. PCCPs are becoming more common, particularly in high-volume AI indications like radiology.

ⓘ This article is for informational purposes only and does not constitute regulatory or legal advice. AIFDA Intel is an independent platform and is not affiliated with the U.S. Food and Drug Administration. Consult a qualified regulatory affairs professional before making regulatory decisions.