Health Canada recently issued a Draft guidance: Pre-market guidance for machine learning-enabled medical devices.
Machine learning (ML) is a subsection of artificial intelligence (AI) which allows training algorithms to establish ML models when applied to data rather than models that are explicitly programmed. Approaches using ML are becoming more and more common. We can think about the automotive industry, robotics, finances, and of course medicine.
In the health sector, the use of ML models can enable much faster illness detection and diagnosis by identifying new observations and patterns in humans. Medical devices (incl. software) that use, in part or in whole, ML to achieve their intended medical purpose are known as machine learning-enabled medical devices ("MLMDs"). The most striking advantage of MLMDs lies in their ability to continue learning as additional data becomes available (including real-world-evidence), to improve the quality of healthcare.
One of the key components of Health Canada's new guidance framework focuses on transparency and the communication of information to caregivers regarding security hazards and device effectiveness to allow them to make informed decisions for their patients.
The draft guidance presents the concept of a predetermined change control plan ("PCCP") that provides a mechanism enabling Health Canada to address cases where the regulatory pre-authorization of planned changes to ML systems is needed to address known risks. For as MLMDs, the PCCP is a fundamental part of the device design. The PCCP should be risk-based and supported by evidence, take a total product lifecycle perspective and provide a high degree of transparency.
Good Machine Learning Practice for Medical Device Development: Guiding Principles
To achieve and maintain the best practice and consensus standards, the U.S. Food and Drug Administration (FDA), Health Canada, and the United Kingdom's Medicines and Healthcare products Regulatory Agency (MHRA) have jointly identified several factors that should be considered from the early development stage. These guiding principles are important for any businesses that design or wish to partner with emerging companies that develop as MLMDs. The regulatory agencies hope that these general principles will empower stakeholders to advance responsible innovations. In practice, we expect that complying with these key principles should help industry participants with the regulatory approval process in Canada and abroad.
The Good Machine Learning Practice for Medical Device Development: Guiding Principles ("GMLP") include 10 guiding principles:
- Multi-Disciplinary Expertise Is Leveraged Throughout the Total Product Life Cycle
- Good Software Engineering and Security Practices Are Implemented
- Clinical Study Participants and Data Sets Are Representative of the Intended Patient Population
- Training Data Sets Are Independent of Test Sets
- Selected Reference Datasets Are Based Upon Best Available Methods
- Model Design Is Tailored to the Available Data and Reflects the Intended Use of the Device
- Focus Is Placed on the Performance of the Human-AI Team
- Testing Demonstrates Device Performance During Clinically Relevant Conditions
- Users Are Provided Clear, Essential Information
- Deployed Models Are Monitored for Performance and Re-training Risks Are Managed
The GMLP are best practices that are intended to evolve with advancements in the field of ML. Stakeholders are invited to provide feedback.
The Fasken team is following all regulatory updates in the Canadian medical device landscape. Do not hesitate to contact one of our members to know more about the new guidance from Health Canada.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.
