Understanding Pre-determined Change Control Plan (PCCP) for AI/ML Medical Devices
20 Jun 2025
Medical devices in themselves are a marvel of modern science. However, as our technologies become more advanced, we tend to upgrade our gadgets and devices as well. This also holds true for medical devices which now largely incorporate Artificial Intelligence and Machine Learning algorithms. And like every development in the medical device industry, this development also needed to be regulated.
By taking this need into consideration, the US FDA published a detailed draft in April 2023 called
“Marketing Submission Recommendations for a Predetermined Change Control Plan for Artificial Intelligence/Machine Learning (AI/ML)-Enabled Device Software Functions.”
In the FDA’s own words:
“This draft guidance describes an approach that would often be least burdensome and would support the ability to modify an ML-DSF while continuing to provide a reasonable assurance of safety and effectiveness across relevant patient populations.“
Basically, this draft is meant for medical device manufacturers who produce/develop/work on medical devices with AI/ML features. In this blog, we will see what the Pre-Determined Change Control Plan (PDCCP) actually means and why it is important for medical devices from a regulatory standpoint.
The Need for Publishing PCCP Draft
Nobody can deny the importance of incorporating Artificial Intelligence and Machine Learning. Usually, the most prominent way to train the AI and ML models is help them learn from the data that is being fed into them. These models then use this training data to produce outcomes. If the data is correct, then the outcomes would be accurate, but if the data is faulty, the outcomes would be inaccurate (and potentially dangerous too).
We can understand this by an example of a diagnostic tool. The said tool can gradually become better at detecting a specific type of cancer by being fed the images of that tumor again and again.
However, each time a change is encountered in the ML-DSF (Machine Learning Enabled Device Software Solutions) of a device, it is considered as a change within the device. With such changes taking place now and then, the FDA can ask manufacturers to submit PMA supplement, Special 510(k), and De Novo Submission (all depending on the device’s regulatory pathway).
“The burden of repeated submissions for iterative modifications can hinder innovation and delay access to beneficial technology. The PCCP aims to address this by allowing FDA pre-authorization of specific future modifications.”
— FDA Draft Guidance, 2023
This constant submission and re-submission of documents can be a source of headache for medical device manufacturers and to avoid it, the FDA requires them to submit a Pte-determined Change Control Plan.
Understanding the PCCP
The PCCP is a document describing
- The modifications required to be made to the software that uses ML, and
- How the manufacturer analyzes those modifications
Regardless of the pathway of medical device, the PCCP would be the standalone section in the manufacturer’s marketing submission.
According to the FDA: “A PCCP that is agreed upon and included in a marketing submission would allow the manufacturer to implement certain changes without submitting a new marketing application each time.”
— FDA Draft Guidance, 2023
Because of the ML component, the PCCP allows manufacturers to form boundaries around any changes within the ML-DSF. By creating such boundaries, there won’t be any need to submit a new marketing submission whenever a recent change is made.
Another way PCCP is beneficial is that it allows medical device manufacturers with ML-DSF to avoid any errors. It also removes any unnecessary regulatory hurdles that can get into the way of marketing the medical devices.
Components of PCCP
There are three major components of PCCP:
- Modification protocol
- Description of modifications
- Impact assessment
Modification Protocol
The modification protocol describes the methods to develop, validate, and implement the modifications in the Description of Modifications section. It will also include the verification and validation activities. The draft guidance of PCCP issued by the FDA has four basic components of Modification Protocol that must be addressed by the manufacturers:
- Update procedures
- Re-training practices
- Performance evaluation
- Data management practices
The FDA emphasizes that the protocol should “describe the methodology for implementing updates, including change control processes and data re-training requirements in a manner consistent with a robust QMS.”
— FDA Draft Guidance, 2023
Description of Modification
In this section, a description regarding planned modifications to ML-DSF is planned. But there will also be a need to add a detailed description of modifications being made to the device’s characteristics and performance.
The FDA asks manufacturers to limit the number of changes to ensure that the changes made are specific and verifiable. The changes being made should be validated within the existing quality system of the manufacturer.
While creating their description of modification, the manufacturers should ask the following questions to themselves:
- Will the implementation of modifications be uniform for every device on the market?
- Will the modifications happen locally according to specific characteristics of a patient or a clinical site?
- Will the modification happen automatically by a software?
- Will the modification be manually done by a human?
It is also important to note that, according to the FDA,
“All modifications included in a PCCP must maintain the device within the device’s intended use. At this time, FDA expects that modifications included in a PCCP should also maintain the device within the device’s indications for use.”— FDA Draft Guidance, 2023
Impact Assessment
Impact assessment is the assessment of risks and benefits of implementing a PCCP for ML-DSF. The resolution of these risks is also included the impact assessment.
As per the FDA, having a robust QMS matters a lot while performing an impact assessment. According to the FDA:
“the manufacturer’s existing quality system should be used as the framework in which to conduct an Impact Assessment for the modifications set forth in the PCCP.”
Which Items Should be Discussed by the Impact Assessment?
According to the FDA, the impact assessment should discuss the following five items:
- Comparison of each modified version of the device with its original version.
- Benefits and risks of each modification, including risks of social harm.
- The way to continue activities in the Modification Protocol so that safety and effectiveness of the device can be ensured.
- How would implementing one modification would affect the other.
- The overall effect of implementing all the proposed modifications.
Want to Implement AI and ML Properly in your Medical Devices?
With the rapid advancements in science and technology, you would not like to be left behind as a medical device manufacturer. However, what remains constant in this situation is the use of QMS.
“FDA encourages manufacturers to adopt a Total Product Lifecycle approach to software development, including use of an appropriate QMS to manage modifications.”
Having a purpose-built QMS that is meant for the medical device industry and is compliant to all the major regulations and standards would be of a great help to you as a medical device manufacturer. At TSQ&E, we can help you achieve this milestone. You can book a session with our medical devices experts. Together, they will help you become compliant with all the major requirements expected from a well-known medical device manufacturer, including the ones related to incorporating AI/ML.
