Integrating AI with 3D printing has the potential to transform the medical device industry and accelerate the development of innovative devices.
Jack Heslin, VP Sales North America at Creatz3D, spoke with MD+DI about the current state of AI and 3D printing integration, how leading companies can bridge the gap between AI expertise and traditional engineering knowledge, and potential pitfalls of adopting these technologies.
At MD&M South in Charlotte, North Carolina, in April, Heslin will explore how AI-driven generative design creates optimized device geometries while machine learning prevents print failures, slashing development time and costs. He'll discuss how leading medtech companies are leveraging this convergence across the entire additive manufacturing workflow, practical implementation strategies, regulatory considerations, and how organizations can position themselves to capitalize on these emerging technologies.
Are there certain device categories or clinical applications where AI + 3D printing integration is delivering the most dramatic results right now?
Heslin: Here is what I think will happen. It's unproven, but it’s a credible speculation. I think in the invasive world things will change. Invasive to you and invasive to me are different things; we are different bodies. People are different shapes and sizes. I think disposability in the medical device, combined with a type of instrument, some kind of tip or invasive part that is unique to each of us. That can come from a body or cavity scan that can go into AI generative design platforms, and the AI will come back and say ‘make the part this way for this unique anatomy.’ That may not be a year away, or 10 years away, but it could happen and will make for a more efficient procedure that takes less time, and has less chance of infection.
How are leading companies structuring their teams to bridge the gap between AI/data science expertise and traditional medical device engineering knowledge?
Heslin: Even though AI is still kind of new in manufacturing, 3D printing isn’t new. That technology began in the 80s and 90s and has evolved. Even with that enormous body of data, most companies still do design work geared toward traditional manufacturing. They think in terms of large volumes, which are cost efficient to produce. It can happen where an AI-generated design cannot be interpreted into a physical product. It exists beautifully in a 2D environment, but something about the geometry, the interior cavities to the device, are not compatible with slicing software. You have a CAD file and a 3D printer, and the slicing software takes your CAD file and writes the G-Code to produce a physical part.
First, we have to create AI-generated designs that regardless of manufacturing technology, are makeable in the first place. To do that, we have to have an AI-generative platform with the knowledge to say ‘you cannot design a part like this. It cannot be made.’ I think it'll be a challenge for medical device companies to educate their designers and engineers until these tools are reliable enough to consistently produce manufacturable designs. That learning curve, we are still at the bottom of it. It's not as simple as ‘Our AI platform says the device should look like this.’ It’s not always makeable. There’s a murky area that still needs to be defined.
What are the unique regulatory considerations when AI is involved in the design or manufacturing process for 3D-printed medical devices, and how are companies documenting AI decision-making for FDA submissions?
Heslin: It's all still toe out the door. It's not just FDA approvals, but it’s insurance companies as well. And that can take years. Typically, it takes reams of data to show the safety of a new procedure, a new drug, a new device. To get reams of data, you need a few thousand hours of actual procedure and test results. That has to be funded. Grant money has to be applied for, companies have to put money aside in their R&D budget to do those tests, to present data to FDA and their insurance provider, and say this is safe. We are at the very beginning of this and there cannot possibly be enough data yet. The question is a bit premature. Illinois politician Dick Durbin said two years ago one of the most honest things I have ever heard anyone in government say. He said, “How are we supposed to legislate something none of us understand?” How will FDA act? There is not enough data yet to make a judgment like that.
Where do you see the biggest risks or pitfalls for companies rushing to adopt AI + 3D printing without proper validation or understanding of the technology limitations?
Heslin: I would say lawsuits. This applies to all AI applications. One of the huge risks is people think they are off the hook for thinking for themselves. “Well the AI said this.” AI is based on statistical probabilities, not certainties. If it’s based on probabilities, then there is always the probability that it will make a mistake. The big risk is, is it too soon to be relying on this technology? Perhaps it's not too soon, but as long as AI is based on probabilities—even if it's 99.99999% accurate—there's still a chance it can be wrong. That is still true today with ChatGPT, Claude, Gemini, and all the rest. I think that's the risk.
Is there anything else you would like to expand on?
Heslin: This is all brand new. This discussion will not be about certainty. There are no definitives. These are the questions, the potentialities, and a lot of good can come out of AI. We're still just dipping our toes in the water.
