E36: How can AI-augmented medical devices be made safer to use?
Updated: Feb 5
The Ask AI check-in with David Boudreau, Director General of the Medical Devices Directorate at Health Canada
Special thanks to Microsoft Canada and Cinchy, the Dataware platform for sponsoring the 4th series of the Ask AI podcast. Click here to subscribe to our podcast and newsletter or search "Ask AI Podcast" wherever you get your shows. All of our episodes and 100s of AI profiles are also available from our chatbot, generously sponsored by Toronto’s Ada Support.
In 2021, the U.S. Food and Drug Administration (FDA), the United Kingdom's Medicines and Healthcare products Regulatory Agency (MHRA), and Health Canada jointly identified 10 guiding principles for Good Machine Learning Practice (GMLP) intended to promote the development of safe, effective, and high-quality medical devices that use artificial intelligence and machine learning technologies.
Tune in to this Ask AI episode as our host Carolyne Pelletier checks in with David Boudreau, Director General of the Medical Devices Directorate at Health Canada to ask the question:
How can AI-augmented medical devices be made safer to use?
Stream the video
Stream on YouTube (don't forget to click 'Subscribe')
Download the podcast
Everywhere else: just search "Ask AI"
Don't forget to follow us to get alerted to future episodes 🤗
Talk to our bot
This check-in and hundreds of AI resources are available from the Ask AI chatbot. Special thanks to our chatbot partner Ada for providing this World-class technology!
David's email (published on request): email@example.com
Statement: Good Machine Learning Practice for Medical Device Development
Subscribe to Ask AI
Click here to subscribe to our podcast and newsletter
Request Team Check-In
If you have news about an investment or project that you’d like us to feature on our podcast, YouTube channel, and newsletter, please feel free to request a Team Check-In.
Sponsors: Microsoft Canada's Free AI Business School and Cinchy, the Dataware platform that eliminates integration Chatbot partner: Ada Support Series Producer: Chris McLellan Episode Host: Carolyne Pelletier Series Editor: James Fajardo Original music: Mike LeTourneau
About Ask AI
Ask AI is a volunteer project exploring Canada's fast-emerging artificial intelligence sector. Our projects include thought leader interviews, team check-ins, and live events. To learn more about sponsorships, guest blogs, and volunteering, please contact us or ask our chatbot.
Please note: this transcript was generated by an artificial intelligence and some typos are invevitable:
Carolyne: Hi, this is Carolyne Pelletier your host for the Ask AI team check-ins. Much like we have collaborations in academia in today's teen check-in.
We're going to hear how Health Canada, the US FDA, and UK's MHRA have come together to identify 10 guiding principles that can inform the development of good machine learning practice.
The goal of these principles as to help promote safe, effective, and high quality medical devices that use AI and ML. And so today we have David Boudreau Director General of the Medical Devices Directorate at Health Canada to talk about this joint statement and more. Stay tuned.
Hi, this is Carolyne Pelletier, your Ask AI team check-in host.
David can you tell me who you are and a bit about the organization that you represent?
David: Yes, of course. So my name is David Boudreau. I'm the Director General of the Medical Devices Directorate at Health Canada.
And Health Canada is the federal government department that regulates the sale and importation of medical devices in Canada.
And so the mandate really here is to ensure that medical devices licensed for sale in Canada meets all the required standards for safety, effectiveness, and quality.
Carolyne: Wonderful. Thanks so much for that introduction. And can you let us know what's new at Health Canada?
David: Yes of course. So I guess I could talk about artificial intelligence and machine learning. I think this is a hot topic. Also called AI ML.
And so these terms are described - or used to describe how computers can learn from data.
So in the field of medical devices, it's certainly becoming an increasing trend which has the potential to transform the healthcare.
And that is because it enables really certain types of medical devices to actually improve on their own performance.
So the department is, I would say an internationally recognized regulator - or a recognized leader, I should say, in this space.
And so we recently issued a joint statement together with the US FDA, which is the United States Food and Drug Administration, as well as the UK in Great Britain.
So the Medicines and Healthcare Products Regulatory Agencies also called MHRH. So together with both the USFDA and the UK MHRA we've issued this joint statement, outlining 10 guiding principles for good machine learning practices also called GMLP.
And this is to promote safe, effective, and high quality medical devices. And I can also share that these guiding principles are intended to help in the development of good machine learning practices around the world and release so that we have a bit of a consistent approach, you know, from regulator to regulator.
And which will then in turn to help companies in the healthcare system. And we also hope to use technologies that were developed in other sectors to benefit medical devices.
Carolyne: Okay. Thanks David and why does this joint statement matter?
David: Well, I guess the ultimate goal of these principles is to improve on the key points of health products, including medical devices which are safety, effectiveness, and quality.
And so if medical devices can learn from the data they receive in a real-world setting and improve themselves in these key areas, then ultimately they will do a better job at improving the health of people who rely on them.
So that's really I guess why it matters.
And for instance, so let's say we consider a software product that has been trained to detect cancer on medical images.
So after deployments once this is getting distributed, it's on the market, the device will perform and will improve - could improve by learning under new data it receives.
And so this would lead to increased benefits to patients.
And that is because it will continue to perform better, building on its own performance. So, I'd like to also say that, you know, it all starts with the manufacturers.
So these new guiding principles will provide the guidance to manufacturers as they are developing their algorithms to operate the devices.
And eventually this will also help companies when they come to me and my team, at the medical devices directorate when they applied for a license, because this is going to provide them with the understanding they need to provide me and my team with a comprehensive submission.
And in turn, this should speed up, you know, the time to market which is extremely key to manufacturers and as well to patients.
So this is a really good example of this joint statement, as you know, in terms of international alignment. And especially in this context with regards to what constitutes good learning.
You know, good algorithm practices and development, and we believe that AI/ML medical devices, you know, as this will evolve and GMLP will evolve and best practices more more broadly.
This will really have an impact on, you know, in medical devices and patients in general.
Carolyne: All right. Great. Thank you so much for that. And who's involved in this joint statement?
David: Sure. So, as I mentioned, the two key players in the context of this joint statement where the US FDA and the UK MHRA. And so we're really pleased actually about this work with these two truly respected and trusted regulators.
And because we were able to really work together on an emerging space. And so the department has been working our tool.
So with the support of these internal and external partners to develop policies and procedures and ensuring, you know, effective regulations of these you know, emerging or innovative devices.
Carolyne: Great. Thanks, David. And can you tell us what's next at healthcare?
David: Sure, so Health Canada is certainly an active participant in the international space of AI/ML.
Actually there is a forum called the International Medical Device Regulators Forum also called IMDRF. And there is a working group there called the AI Working Group and there is also another organization that you certainly know, the World Health Organization, WHO.
And they have The International Daily Communication Union group.
And they have a focus group there called AI for Health, and then with the number four health. And so I'm the IMDRF AI working group is currently - and where we have an active role, Health Canada - is currently working on the terms and definitions that relate to machine enabled medical devices.
We'd like to basically establish this collectively with our trusted regulators and we believe that this joint statement that we were just talking about with the USFDA in the UK MHRA that we've developed will certainly inform or help inform the development of future international work, including the one with the IMDRF that I was just talking about.
And the department is also working on additional guidance.
So specifically to department, we're deploying our own guidance document to further help manufacturers of medical devices to understand, you know, what needs to be submitted in the context of the submission for medical devices including AI/ML component.
And this is going to really help them and guide them for the requirements for safety, effectiveness, and quality.
And we're expecting to go out with our public consultation for this new guidance on AI/ML for med devices in 2022. So it's coming really shortly. So this is really what's next for us.
Carolyne: Great, David. And where can we get more information on this topic?
David: Well, you can certainly get more information on our website. So we did publish this together with the US FDA and MHRA so you can have access to this joint statement and I can provide you with the link.
And also, you know, I like to make myself available, should people have questions, who'd like to engage with me on this topic and I can also provide you with my email address.
Carolyne: Perfect. Thank you so much for joining us today.
David: My pleasure.