Article

Predetermined Change Control Plans (PCCPs): A change management solution for AI in healthcare

Published on January 14, 2025 | 6 min read
predetermined-change-control-plans

Key takeaways

  • The increasingly rapid adoption of digital health technology brings with it new regulatory complexities
  • Current change management models for regulation are outdated and inflexible
  • Predetermined Change Control Plans offer a better solution to manage post-market modifications of medical devices

Software as a medical device (SaMD), a tool that more and more often leverages artificial intelligence (AI) technology, already plays a large part in healthcare.1 Existing technology is being used for image analysis, signal processing (e.g. from electrocardiograms), and to provide data to support clinical decision-making.2 It is clear that digital technology is fundamentally transforming the healthcare landscape, but setting the course for a smooth transition requires foresight and regulatory mechanisms that can keep up with the rapid pace of technological change.

AI technology learns from data and can improve its performance over time. This evolution inevitably leads to the need for regular software updates. For consumer products like smartphones, software updates are a relatively simple, albeit at times inconvenient, undertaking. However, when it comes to digital health products, where patient safety is paramount, regulatory considerations make software updates quite a bit more complicated.

The regulation of AI in healthcare is still nascent, and not yet fully developed in a way that can bring the full benefits of AI to patients. Current regulatory models are designed for ‘locked’ healthcare solutions, therefore a new model of change management is required that can account for flexible and evolving digital technologies.3 Predetermined Change Control Plans (PCCPs) are a regulatory concept proposed to better manage updates to on-market products. PCCPs offer a pragmatic approach by allowing regulators to pre-authorize intended modifications within strict parameters, to ensure products remain safe and true to their intended use. While PCCPs can assist medical devices and IVDs more broadly, this discussion focuses on the benefits for SaMD in particular.

Why is a new approach needed?

Section textaCurrent change management requirements in medical device regulation exist to protect patients across a device’s entire product life cycle. In nearly all cases, regulations require that all significant, substantial, or major changes to a medical device be submitted prior to implementation to regulatory authorities, as with, for example, the US Food and Drug Administration (FDA). 4  These processes are intended to protect patients, ensuring that the device continues to function after modification. This typically leads to a lengthy re-authorization process, burdensome on both manufacturers and regulators.

Drawn-out regulatory processes are not compatible with AI models that are trained on data, and must change over time. For instance, a model’s performance can decline significantly if the training data no longer represents real-world deployment. This is called data drift.5 The ability to respond quickly to data drift is critical for AI-enabled medical devices to perform at their best. Another benefit of PCCPs is that they unlock the ability of models to be tailored to fit specific circumstances, populations, or locations.6

The current time-intensive regulatory process presents a barrier to these changes, and can often delay manufacturers from improving their device or responding to changes in the device or environment. This situation threatens to stifle the adoption and innovation of devices for patients.3  In response, the international medical device community is discussing the use of PCCPs as a different way of managing change for medical devices utilizing AI technology.

What are Predetermined Change Control Plans (PCCPs)?

A PCCP is the documentation that outlines a set of future modifications intended to be made to the device. A PCCP allows manufacturers to action changes within the scope of the documentation without seeking further authorization, so long as these changes do not change the intended use or purpose of the device.

Draft guidance published by the US Food and Drug Administration (FDA) in August 2024 outlines three key elements that should be included in PCCPs:7

  1. Description of Modifications: a list of proposed modifications as well as the rationales for these changes.
  2. Modification Protocol: a description of the methods that will be followed when developing, validating, and implementing the modifications listed.
  3. Impact Assessment: an assessment of the benefits and risks of the modifications as well as mitigation measures.

PCCPs allow for the agile modifications that are essential for the maintenance of performance in SaMD (as well as other medical devices / IVDs), while still providing a comparable level of safety assurance as in current change management models.8

How can PCCPs enable innovation?

The benefits of using PCCPs are obvious for all those involved with the medical device process. By allowing for modifications within the bounds of the PCCP, manufacturers can swiftly retrain their models to guard against data drift, and to better suit different environments. The process also encourages manufacturers to proactively plan ahead, predict, and anticipate future changes in the post-market phase, thus providing more certainty and safety in devices.

As an authorized PCCP no longer needs to be reviewed each time a change is required, there is a reduced burden on regulators. In turn, this frees up time to devote to activities that are more directly related to safety.

For patients, innovative devices that need frequent modifications to maintain and improve performance now have a path to market. Patients gain quicker access to these technologies, potentially improving patient care.

predetermined-change-control-plans

What does the future hold?

As with any new field in healthcare, there are likely to be significant regulatory questions and challenges, particularly so in an area as dynamic and complex as SaMD. However, as the use of SaMD increases in scale, it is essential that a practical solution is implemented to keep up with demand.

Consideration must be given as to how to maximize the value of PCCPs, including:

  • International alignment: If PCCPs are highly internationally aligned, their value is maximized. If they are highly divergent, their value is diminished as PCCPs would become unwieldy if there are too many different rules in different countries. This is particularly important for manufacturers with global portfolios.
  • Integration: A clear idea of how PCCPs will integrate and interact with existing regulatory documentation, such as quality management software and risk management systems, is required.
  • Flexibility: If PCCPs are to be a better form of change management, they must deliver value to patients and have a unique selling point. They must be sufficiently flexible, minimally burdensome, and require a set of documentation that is reasonable to provide in the premarket phase.
  • Breadth: PCCPs, while they have special use in the context of AI, have broader applicability beyond SaMD and can be used for medical devices and IVDs.

Efforts are being made to harmonize approaches to PCCPs internationally between the FDA in the Unites States, Health Canada, and the Medicines and Healthcare products Regulatory Agency in the UK.9 Additionally, there is also work underway within the International Medical Device Regulators Forum for early alignment on PCCPs that includes many different jurisdictions.

There is still a hill to climb to harmonize regulatory approaches and mechanisms in the area of SaMD. However, these challenges must be overcome if SaMD manufacturers are to have the clarity and confidence necessary to bring innovations to market.3 With a concerted effort, and with smart approaches like PCCPS, healthcare manufacturers and regulators can work together to bring life-changing technologies to patients worldwide.

Person facing winding roads to a flag

Accelerating innovation with PCCP

Accelerating innovation with Predetermined Change Control Plans

Get our latest insights

Join our community and stay up to date with the latest laboratory innovations and insights.

Contributor

Johan Ordish headshot

Johan Ordish, MA, MA (Cantab)

Head of Digital Health and Innovation Policy, Roche Diagnostics

Johan Ordish is the Head of Digital Health and Innovation Policy at Roche Diagnostics, where he helps prepare the company for incoming regulation that applies to digital health products. Johan was previously Head of Software and AI at the UK Medicines and Healthcare products Regulatory Agency (MHRA), regulating all medical device software for the country. He is also an Honorary Associate Professor at the University of Birmingham and a By-Fellow at Hughes Hall, University of Cambridge.

Newsletter for healthcare leaders and experts

Written for experts by experts, we offer the healthcare newsletter of choice when it comes to leading healthcare transformation.

Healthcare Transformers delivers insights on digital health, patient experience, healthcare business, value-based care, and data privacy and security—key topics and emerging trends facing healthcare leaders today. Collaborating with esteemed industry experts and innovators worldwide, we offer content that helps you gain first-hand knowledge, explore challenges, and think through solutions on the most pressing developments and issues. Subscribe to our Healthcare Transformers newsletter today and get critical discussions and invaluable perspectives delivered straight to your inbox.

References

  1. Medicines & Healthcare products Regulatory Agency. (2024). Paper available from https://www.gov.uk/government/publications/software-and-artificial-intelligence-ai-as-a-medical-device/software-and-artificial-intelligence-ai-as-a-medical-device [Accessed December 2024]
  2. DLRC Group. (2023). Paper available from https://www.dlrcgroup.com/wp-content/uploads/2023/12/Whitepaper-Revolutionising-Healthcare-Unleashing-the-Power-of-AI-in-Medical-Devices.pdf [Accessed December 2024]
  3. Ernst & Young. (2022). Paper available from https://www.ey.com/en_gl/insights/law/how-the-challenge-of-regulating-ai-in-healthcare-is-escalating [Accessed December 2024]
  4. U.S. Food & Drug Administration. (2017). Available from: https://www.fda.gov/media/99785/download [Accessed December 2024]
  5. Berkman S. et. Al. (2023). British Journal of Radiology, 96,1150. Paper available from: https://academic.oup.com/bjr/article/96/1150/20220878/7499000     
  6. Futoma J et al. (2020) The Lancet Digital Health, 2, 9, e489 – e492. Available from: https://www.thelancet.com/journals/landig/article/PIIS2589-7500(20)30186-2/fulltext
  7. U.S. Food & Drug Administration. (2024). Available from https://www.fda.gov/regulatory-information/search-fda-guidance-documents/predetermined-change-control-plans-medical-devices [Accessed December 2024]
  8. U.S. Food & Drug Administration. (2024). Available from: chrome-extension://efaidnbmnnnibpcajpcglclefindmkaj/https://www.fda.gov/media/166704/download [Accessed December 2024]
  9. Medicines and Healthcare Products Regulatory Agency UK. (2023). Available from: https://www.gov.uk/government/publications/predetermined-change-control-plans-for-machine-learning-enabled-medical-devices-guiding-principles  [Accessed December 2024]