Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsAn AI/ML device software function’s predetermined change control plan (PCCP) is audit-ready when every planned change can be traced from its authorized scope through data, testing, risk decisions, release, user communication, and postmarket monitoring. No plan can guarantee an audit outcome: FDA’s August 2025 final guidance gives recommendations alongside applicable legal requirements, and an authorized PCCP covers only changes implemented within its boundaries.
What an FDA-authorized PCCP does—and does not do
FDA’s final guidance, Marketing Submission Recommendations for a Predetermined Change Control Plan for Artificial Intelligence-Enabled Device Software Functions, dated August 2025, addresses AI-enabled device software functions (AI-DSFs), including device constituent parts of device-led combination products reviewed through 510(k), De Novo, and PMA pathways. FDA recommends including a PCCP in a marketing submission so that specified changes implemented consistently with the authorized plan do not each require a new marketing submission.
As an Amazon Associate I earn from qualifying purchases.
A PCCP is therefore not an internal change log or blanket permission to modify a device. It must first be established through marketing authorization, and a change is covered only if it stays within the authorized plan and applicable requirements. FDA guidance is generally nonbinding; the document explains that “should” means suggested or recommended unless a statutory or regulatory requirement applies.
The joint FDA, Health Canada, and MHRA guiding principles offer a useful design lens: keep the plan focused and bounded, risk-based, evidence-based, transparent, and oriented to the total product lifecycle. They emphasize limiting planned changes to the original intended use or purpose, assessing individual and cumulative changes, using scientifically and clinically justified performance measures, and maintaining transparency before and after modifications. The U.S. pathway does not automatically govern other markets; each jurisdiction may implement PCCPs differently.
#1 Best Overall
What belongs in the three linked PCCP sections?
FDA identifies three core sections: Description of Modifications, Modification Protocol, and Impact Assessment. They should work as one chain: define the allowed change, prescribe how it will be developed and evaluated, then explain why the controls address the risks it creates.
1. Description of Modifications: define the boundary
Describe the specific changes the manufacturer proposes to implement and the modified AI-DSF’s characteristics and expected performance. Specify relevant intended uses, populations, and environments where applicable. Make the boundary clear enough that a reviewer can distinguish a planned change from one that is out of scope.
Changes under a PCCP must maintain intended use and, where applicable, substantial equivalence; FDA generally expects the device to remain within its indications for use. Because prospectively assessing many indication changes may be difficult, FDA encourages discussion with the review division through Q-Submission when the boundary is uncertain.
2. Modification Protocol: make the process executable
Write the protocol as a step-by-step procedure, not a promise to “revalidate” later. FDA identifies data management, retraining, performance evaluation, and update procedures as key elements. For each planned modification, specify:
- Data management: data sources; inclusion and exclusion rules; provenance; representativeness; and how training, tuning, and test data are separated.
- Retraining: triggers, methods, and controls for overfitting and bias.
- Performance evaluation: what triggers testing; independent, sequestered test data; clinically justified metrics; statistical analysis; sample-size rationale; relevant high-risk populations and challenging cases; and predefined acceptance criteria.
- Device-level assessment: how safety and effectiveness will be evaluated across the device, including whether an improvement in one metric could cause unacceptable degradation elsewhere.
Specify what happens if a test fails. FDA recommends recording an unresolvable failure—for example, failure to meet predefined criteria—and not implementing that modification. If root-cause analysis shows the cause is unrelated to aspects of the PCCP, it may justify repeating testing; document the rationale rather than rerunning tests silently until they pass.
Define the release path as well: whether implementation is automatic or manual, software and version controls, release authorization, rollback or stop mechanisms, user communication, updated instructions or training where appropriate, and post-implementation surveillance. FDA expects manufacturers to monitor safety and effectiveness over time and recommends procedures for communicating relevant changes.
Rank #3
3. Impact Assessment: connect changes to risk controls
For each modification, connect expected benefits and hazards to risks, mitigations, and residual risk. Consider effects on the AI-DSF and the overall device, including other software functions or hardware where relevant. Assess individual and cumulative changes, and explain why the protocol’s evidence is sufficient to support safe and effective performance across the intended populations and environments. The assessment should show how each control addresses the risks introduced by the planned changes.
Free tools Windows power users keep installed
One-click scans. No signup required.
How to make the evidence trail reconstructable
FDA recommends presenting the PCCP as a standalone section in the marketing submission, with a title and version number, prominent references in the cover letter and table of contents, and cross-references from other sections that discuss PCCP content. If the plan changes during review, submit a clean final version with a current version number. The authorized plan is referenced in the device’s authorization letter.
For each implementation, retain versioned records sufficient for an independent reviewer to reconstruct the decision and execution. A practical record set includes:
- The authorized PCCP and version, change request, rationale, and scope determination.
- Data provenance and partitioning records, plus training, tuning, and test records.
- The protocol version and execution record, metric results, acceptance decisions, subgroup and challenging-case results.
- The impact assessment, risk controls, and residual-risk rationale.
- Failures, deviations, root-cause analyses, and rationale for any retesting.
- Implementation and release records, rollback readiness, labeling, instructions, training, user notices, and postmarket safety and performance monitoring.
This record set is a practical synthesis of FDA’s recommendations, not a separately published FDA audit checklist. The point is traceability: the authorized scope, evidence, decision, and real-world implementation should connect without relying on undocumented judgment.
What should users be told?
FDA recommends that labeling inform users that the device uses machine learning and has an authorized PCCP, and that updates may affect performance, inputs, or use. As applicable, updated labeling can describe implemented modifications, current performance, relevant training, tuning, and test data, inputs and outputs, validation requirements and evidence, how changes were implemented, and how users are informed—for example, through updated instructions or version history.
What if a change falls outside the plan or fails testing?
A failed planned change should not be released merely because the PCCP lists it. If the failure is unresolvable under the protocol, record it and do not implement that modification. Where a documented root-cause analysis supports repeating a test, preserve the analysis and retest rationale.
If the manufacturer cannot follow the authorized plan, or a proposed change falls outside its boundaries, do not represent that change as covered. FDA calls failure to follow or inability to follow the authorized plan a deviation; the guidance notes that significant out-of-plan modifications likely require a new marketing submission before implementation. A manufacturer may choose not to use an authorized PCCP or may submit a new marketing submission instead.
How to judge whether a PCCP is sufficiently bounded and evidenced
There is no single FDA-endorsed template or universal quantitative threshold for all devices. Review the plan against the characteristics of the particular function and its risks:
- Are planned changes specific and bounded, including by intended use and applicable indications?
- Are risks, reversibility, and cumulative effects addressed rather than treating each change in isolation?
- Are data representative of intended populations and environments, with independent evaluation data?
- Are acceptance criteria, statistical methods, and coverage of high-risk groups and challenging cases justified in advance?
- Does the plan explain release controls, user transparency, and ongoing monitoring in addition to model performance?
These questions reflect the dimensions FDA’s guidance and the joint guiding principles emphasize. They help reveal where a plan’s scope or evidence does not yet support the changes it proposes.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
