At COMSOL‘s annual conference this year, a panel of medical device industry members and innovators gathered to discuss computer modeling in the design and development of devices. The panel included Paul Belk, principal engineer at Abbott (NYSE:ABT); Dr. Steven Conrad, professor of medicine at Louisiana State University; Freddy Hansen, senior R&D physicist at Abbott; Nicolas Huc, product manager of the heat transfer module at COMSOL; and moderator Kyle Koppenhoefer, principal at AltaSim Technologies.
Kyle Koppenhoefer: How do you see multiphysics modeling as a strategic differentiator for your technology development processes?
Paul Belk: The interesting thing is the moment you decide you’re going to model at all, you completely modify your design strategy, because I work with some gifted engineers, kind of from the old school, who approach every new project by looking for something they’ve designed recently, making maybe a modification to it, sticking it in, and seeing if it works.
And you can get some distance out of that approach, but of course, if it doesn’t work, it’s, ‘Well, let’s try to guess why.’
On the other hand, if you’re going to model at all, one of the first things you have to do is sit down and start writing down what you think are the important interactions between the different systems in the model, and this forces you into a quantitative design mindset and a quantitative test mindset before you ever push the compute button. So even if your model never runs, which it will, but even if your model never runs, you’ve already designed things very typically just because strategically you’ve forced yourself to say, ‘Let’s understand the mathematical relationships between this.’
Strategically, what’s unique to working with biomedical devices is that you have so little control and understanding of a lot of parameters that make your life very difficult. So you can characterize your device. You can characterize everything you build. You can, to some extent, get material scientists to tell you how this particular thing is going to behave or you can measure it on the bench. But the moment you stick it into the body, it’s going into a very hostile, very unpredictable environment.
There are tons and tons of papers published explaining, “This is what the tissue does and this is what the blood flow does and everything.” But the problem is, lots of papers describe the same thing and they all describe it differently and the reason is that biological systems don’t sit still long enough for you to take good measurements on them, and if you can get a good measurement on them, they then go and change just to demonstrate that you don’t really know what you’re doing. So strategically, you have to create a model because the only thing scarier than modeling a biological device is releasing a biological device without having modeled it.
Steven Conrad: I’d like to echo what Paul said about biological variability. It’s one of the issues that has to be dealt with when trying to model biological systems because you’re trying to validate something in a biological environment. There’s an inherent variability that’s going to need a measurement that’s going to give you a problem. We don’t have that kind of problem in physical systems.
Freddy Hansen: In medical technology, we often do animal studies and we use those if we want to validate our product before we go to clinical studies and one thing we often can’t do in animal studies is to measure everything. But you can maybe measure a couple of things in your animal study, then essentially post-process your animal study results in COMSOL, build the model for it, and get all of the other measurements analyzed and that can actually help you decide between two different products, which way to go when a measurement is possible.
Belk: Yes, especially when you’re doing anything implantable. One of the things I didn’t mention was the inherent opacity in biosystems. You stick something into the heart or you stick something into the ear and you can see whether it had the effect that you wanted, but you can’t really stick probes in to measure what’s going on. Simulation, at least lets you do things like visualize the fluid quality. That would be very, very difficult to measure in situ.
This conversation has been condensed and edited for length.