Sponsored By

5 Guiding Principles for Predetermined Change Control Plans in MLMDs

US, UK, and Canada released a joint document identifying principles that draw upon a previous 2021 guidance on good machine learning practices in AI/machine learning-enabled medical devices.

Katie Hobbins

November 6, 2023

4 Min Read
Human_Factors_Guidance_in_Combination_Medical_Devices_66
sdecoret / iStock via Getty Images

The United States’ FDA, Health Canada, and the United Kingdom’s Medicines & Healthcare Products Regulatory Agency recently published five guiding principles for predetermined change control plans (PCCPs) in machine learning-enabled medical devices. This comes after a 2021 joint document identifying 10 guiding principles to inform the development of Good Machine Learning Practice (GMLP).

The newly announced principles build upon the original 2021 guidance — particularly principle 10: Deployed Models Are Monitored for Performance and Re-Training Risks Are Managed, which states, “Deployed models have the capability to be monitored in ‘real world’ use with a focus on maintained or improved safety and performance,” according to FDA. “Additionally, when models are periodically or continually trained after deployment, there are appropriate controls in place to manage risks of overfitting, unintended bias, or degradation of the model (for example, dataset drift) that may impact the safety and performance of the model as it is used by the Human-AI team.”

Advancements in digital health technologies, like artificial intelligence/machine learning-enabled medical devices (MLMD), have regulatory expectations aligned with best practices for development and change management. The quality support that principles such as those outlined in the GMLP Guiding Principles uphold, according to the document, enable patient benefits like earlier access to innovative technologies and more accurate diagnosis.

While change management helps ensure continued safety and effectiveness of devices throughout the total product lifecycle (TPLC), certain changes to MLMDs like to a model or algorithm may be significant enough to require regulatory oversight, such as additional premarket review.

“Such regulatory expectations may not always coincide with the rapid pace of MLMD development,” according to the document.

That’s where the discussion of PCCPs comes into the conversation. Internationally, the medtech industry has investigated the use of PCCPs as a way to manage certain device changes in which regulatory authorization before marketing is typically required. The document specifies that the term PCCP, for their purposes, describes a plan, proposed by a manufacturer, in relation to certain planned modifications to a device, the protocol for implementing and controlling those modifications, and the assessment of impacts from the modifications.

Some examples of when PCCPs could be used include aligning regulatory processes with a rapid and ongoing approach to change management in MLMDs; managing risks in a quick and ongoing manner through monitoring, maintenance, and improved device performance; and upholding regulatory standards to device safety and effectiveness.

Five guiding principles

  1. Focused and Bounded: The first guiding principle describes specific changes a manufacturer intends to implement. Of note, such changes are limited to modifications within the intended purpose of the original MLMD. This can include the “extent of planned changes and scope of the MLMD with changes implemented, plans in place to safely modify the device withing the bounds of the PCCP, like methods for verifying and validating the changes and mechanisms to detect and revert or stop implementation of a change that fails to meet specified performance criteria, and the impacts of the planned changes,” according to the regulatory bodies.

  2. Risk-Based: Intent, design, and implementation of a PCCP driven by a risk-based approach that adheres to the principles of risk management strengthens the value and reliability of the device. This is relevant throughout the TPLC and ensures that individual and cumulative changes remain appropriate over time for the device and use environment.

  3. Evidence-Based: Data generated throughout the TPLC, according to the guide, is important for the ongoing safety and effectiveness of the device with a PCCP, demonstrates that benefits outweigh associated risks, and establishes that risks are adequately managed and controlled.

  4. Transparent: Best practice is to provide clear and appropriate information and detailed plans for ongoing transparency. This helps stakeholders stay informed about device performance and use before and after changes are implemented. Examples for when transparency should be considered include characterization of data used in development and modifications, comprehensive testing for planned changes, and monitoring, detection, and response to deviations in device performance.

  5. Total Product Lifecycle Perspective: Using a PCCP from a TPLC perspective, the regulatory bodies wrote, can “elevate the quality and integrity of a PCCP by continually considering the perspectives of all stakeholders as well as risk management practices throughout the TPLC, and can use and support existing regulatory, quality, and risk management measures throughout the TPLC to ensure device safety by monitoring, reporting and responding to safety concern.”

The document notes that PCCPs may be developed and implemented in different way in different regulatory jurisdictions.

About the Author(s)

Katie Hobbins

Managing Editor, MD+DI

Katie Hobbins is managing editor for MD+DI and joined the team in July 2022. She boasts multiple previous editorial roles in print and multimedia medical journalism, including dermatology, medical aesthetics, and pediatric medicine. She graduated from Cleveland State University in 2018 with a bachelor's degree in journalism and promotional communications. She enjoys yoga, hand embroidery, and anything DIY. You can reach her at [email protected].

Sign up for the QMED & MD+DI Daily newsletter.

You May Also Like