Hong: There are different classifications of AI in medical device technologies.

The first level is computer-aided detection system. The next level is computer-aided diagnostic systems, which is not just flagging a problem, it can tell you about cancer severity, etc. The next level can be computer-aided triaging, and then you have fully automated AI, which in my opinion does not exist.

It will take some time for us to get there [fully automated AI]. Regardless of the form and detail, and how and when we will get there, we need to have a really good understanding of responsible use of AI for patient care. The recently issued Executive Order by the Biden Administration on safe, secure and trustworthy AI captures really well both the potential and also the potential pitfalls for using AI. (Also see "‘We Need To Be Transparent About Our Failures,’ Experts Say About AI Development" - Medtech Insight, 21 November, 2023.)

We as a device manufacturer should work very closely with regulators, clinicians, patient advocacy groups and the scientific community. We need to sit down and understand what level and what specific way of using AI would be needed to enable that level of autonomy.

Nobody has the answers at this point. The journey is just beginning. Once we have good agreement among those groups, then we can talk about the ramifications or how the technology can be used for that type of scenario [autonomous devices]. My personal view on the AI field is that longer-term, AI will be greatly optimized for patient outcome.

The optimization of patient outcome can come in a few different ways. One is for individualized treatment. Another way can be for early detection of diseases or more accurate detection of the diseases. That is the holy grail. That said, I feel we don't have a good understanding of how to regulate that kind of technology yet.

For us to get there, we need good scientific and clinical evidence like lots and lots of outcome studies with the appropriate amount of subjects. All those things do need to happen for us to get there. I don't know how long it will take. It can be five years, 10 years or 20 years. We will get there at some point. (Also see "Medtronic’s Med Surg President Speaks On Development Strategy, Recent Deals And Hopes For Future" - Medtech Insight, 15 November, 2022.)

Medtronic Endoscopy’s AI Head Contemplates ‘Responsible Use Of AI’

More from AI

More from Digital Technologies