Let’s talk about something exciting—and sometimes a little challenging—: the world of artificial intelligence (AI) in medical devices. As an engineer specializing in UX and UI, I’m convinced that great products require both an understanding of the technology and in-depth knowledge of its use and users.
As soon as AI is used in a medical device, it quickly becomes a topic of interest—both from a technological standpoint and from the users’ perspective. In addition, there is a growing number of guidelines (many of which are still in draft form) that govern the use of AI in medical devices.
That’s reason enough to take a closer look at these guidelines. In this series, we’ll be asking: What are the implications for usability and UI design? Each part of the series will examine one guideline in detail.
In this overview, we present the guidelines that we’ve already examined in detail. Think of it as both a summary and a table of contents. Since our series isn’t complete yet, it’s worth checking back periodically to see if any sections have changed or if new ones have been added. Alternatively, you can stay up to date by subscribing to our newsletter.
Let’s go!
Part 1: Good Machine Learning Practice for Medical Device Development – Guiding Principles
The first part of the series focuses on the IMDRF’s Good Machine Learning Practice (GMLP) Guiding Principles. The draft presents 10 principles that serve as a starting point for the development of AI-based medical devices. But what do they mean for usability?
The usability of AI-based medical devices must be systematically integrated from the outset, as regulatory authorities expect these devices to fit seamlessly into clinical workflows and be easy to use. In particular, human-AI collaboration, the transparency of AI decisions, and multidisciplinary development play a central role. In addition, safety aspects—such as error-preventing authentication processes—must be implemented in a user-friendly manner and clearly communicated.
Part 2: Artificial Intelligence-Enabled Device Software Functions: Lifecycle Management and Marketing Submission Recommendations
The second part of the series examines the FDA guidance document “Artificial Intelligence-Enabled Device Software Functions: Lifecycle Management and Marketing Submission Recommendations.” The draft specifies how AI-enabled medical devices must be designed to ensure safety and transparency throughout their entire lifecycle.
Usability is increasingly becoming a focus for regulatory authorities: Users must not only understand the results but also be able to understand how they are arrived at. Human-AI collaboration, risk analysis, and dynamic decision-making must be designed to be transparent and user-friendly. Clear documentation and continuous monitoring are essential to minimize risks such as misunderstandings or erroneous decisions.
What happens next?
As mentioned in the introduction, this series is being updated on an ongoing basis. In doing so, we’ll examine whether the drafts have evolved. We’ll also be incorporating additional guidelines into our series. If there’s a specific guideline you’d like to see covered here, please feel free to contact us.
