Presentation
Applying Human Factors Principles to Development of AI-Based Medical Devices
SessionHealth AI (DH4)
DescriptionThis presentation is a discussion of how to apply human factors in the end-to-end design and development process of AI-based medical devices. The material is drawn from desk research, the author’s own UX research and human factors work on AI-powered digital stethoscopes, and input from the author’s colleagues in UX research, human factors and regulatory colleagues who have also worked on AI-based medical devices. AI is the new hot topic in every industry, and since the disruptive launch of ChatGPT, more and more products are claiming “AI powered” or “now with AI.”
The first section is an overview of AI products in healthcare, starting with AI-powered documentation and radiology tools where adoption seems high, followed by other medical devices where adoption seems slower. The second section discusses whether human factors challenges for AI-based medical devices are new or old. This section includes a definition of AI, emphasizing how AI is an extension and expansion of machine-learning algorithms, which can be thought of as the “next generation” of computer-based algorithms. Then, there is a case study of a product with computer-based algorithms. The case study looks at applying the same tried and tested human factors principles to both products with computer algorithms and products with AI. The third and final section discusses unique human factors challenges in AI-based medical devices. Topics include clinician perceptions / biases, performance variability and operational supervision. Individual clinicians will tend to perceive AI as a threat to their practice of medicine, whereas health system leaders will tend to be more pragmatic and be actively considering how to carefully embrace AI-based technology into their health systems.
In summary, there are two key takeaways for this presentation. The first takeaway is there is a wealth of knowledge from UX research and human factors from the healthcare space that can be tapped into. The second takeaway is an understanding of the unique challenges that need to be overcome, in order to successfully launch an AI-based product in healthcare.
The first section is an overview of AI products in healthcare, starting with AI-powered documentation and radiology tools where adoption seems high, followed by other medical devices where adoption seems slower. The second section discusses whether human factors challenges for AI-based medical devices are new or old. This section includes a definition of AI, emphasizing how AI is an extension and expansion of machine-learning algorithms, which can be thought of as the “next generation” of computer-based algorithms. Then, there is a case study of a product with computer-based algorithms. The case study looks at applying the same tried and tested human factors principles to both products with computer algorithms and products with AI. The third and final section discusses unique human factors challenges in AI-based medical devices. Topics include clinician perceptions / biases, performance variability and operational supervision. Individual clinicians will tend to perceive AI as a threat to their practice of medicine, whereas health system leaders will tend to be more pragmatic and be actively considering how to carefully embrace AI-based technology into their health systems.
In summary, there are two key takeaways for this presentation. The first takeaway is there is a wealth of knowledge from UX research and human factors from the healthcare space that can be tapped into. The second takeaway is an understanding of the unique challenges that need to be overcome, in order to successfully launch an AI-based product in healthcare.
Event Type
Oral Presentations
TimeTuesday, March 268:30am - 9:00am CDT
LocationSalon A-2
Digital Health