Jun 29, 2025 AI is taking center stage in the tech world. Large-language models and AI agents are now gaining traction as productivity-boosters on mainstream platforms such as Google Search and Siri. And the 3D printing industry is closely following suit. The practicality of using AI for medical 3D printing rests on developing trustworthy algorithms that prioritize patient safety while advancing innovation, a harrowing feat considering the black-box nature of many AI algorithms and need for diverse, large-scale datasets. But for the four innovators we heard from at our latest 3DHEALS event, it’s a challenge worth tackling. In this event recap, we take a look at how these changemakers are creating AI tools that collaborate with doctors to predict outcomes before surgery, recommend optimal bioink formulations for digital dentistry and gum disease, create a ChatGPT for operating and servicing 3D printers, and image segmentation for a personalized pessary device. Watch the recording on 3DHEALS Courses.
Cosmetic surgery: using AI to predict the future
Big tech companies are looking at AI agents, or programs that perform actions on the user’s behalf, to expand the role AI plays in our everyday decision-making. For 3D printing innovators, they’re envisioning AI as a key player in augmenting clinical decisions.
William Jung, Business Development Director at FITme, describes how the company is developing “THE FACE ON,” an AI engine capable of segmenting anatomical structures in CT scans and predicting patient outcomes after surgery. Anatomical segmentation is becoming a popular application of AI in the field (just see what Materialise and others are doing). And it’s the patient outcome prediction that’s pushing the boundaries of what AI can do, making it more of an oracle and not just a replacement for manual tasks.
The project is still in the works, but such software could provide patients with side-by-side before and after pictures of their face during the planning process for cosmetic facial surgery. The AI can then recommend the best shapes and sizes for the company’s custom 3D-printed molds that are then used to create silicone implants for the surgery. The goal of this AI assistant is to reduce uncertainties patients may have about their procedure, building trust that the surgery will yield the results they desire.
What’s key in the company’s strategy is that they’re not simply an implant company. Rather, they’re an implant company driven by data. If 3D printing innovators want to truly leverage AI, they’ve first got to invest in collecting and storing as much data as they can get. Techniques that FITme uses, such as Cascade nnU-Net3D (a type of AI model), require large-scale image datasets to prove effective.
Abundant and diverse data are everything, especially because AI products can fail for many reasons: the training data were poor quality, the AI was trained on one patient population but fails when used on a different population, the AI is used at another hospital that uses a different brand of imaging equipment than they were expecting, and more.
AI predicts bioink formulations for bioprinted gum
Getting it just right is no easy feat in 3D printing, especially considering the myriad of printing parameters and bioink formulation combinations that one must experiment with to create the optimal product. But for our speakers, AI is the way out of this conundrum.
Dr. Gopu Sriram, Assistant Professor at the Faculty of Dentistry, National University of Singapore, is harnessing AI to predict which experiments to run next. Developed by colleague Dr. Dean Ho, Professor of Biomedical Engineering at NUS, the “IDentif.AI” platform was designed to help researchers pick which combinations of different drugs would produce the most promising new therapies for various infectious diseases, such as COVID-19. And Dr. Sriram has adapted the technology for bioprinting.
Dr. Sriram and colleagues were able to use the AI platform to predict the output filament diameter of their bioprinter based on the complex combination of pressure, printing speed, and nozzle diameter used. Without AI, they would need to have performed more than a thousand prints just to test four bioink formulations, which would be a costly and time-consuming ordeal.
“Contracting the timeline” is a key accelerator in biopharma research and drug developement.
Published in Advanced Healthcare Materials, they applied the software to bioprinting gum tissue grafts to treat gum recession by first starting with a training set of 25 prints using different printing settings and bioinks. The Identif.AI platform then provided the researchers with recommended printing settings to achieve the filament diameter they desired. Predicting the resulting characteristics of their bioprinted gum tissues is especially important since the size and shape of such grafts can significantly impact treatment outcomes.
AI is opening doors for accelerating the rate of research, acting as a compass for which direction to head in next. However, the critical challenges ahead will be validating these AI systems (which will inevitably take much time and expense), transferring the technology to all the various niche applications of bioprinting, and expanding its ability to “understand” more than just a few types of drugs or bioinks.
Despite these challenges, Dr. Sriram’s work shows that AI’s utility in such a specific area of bioprinting (i.e., gum tissue grafts) is not just promising, it’s already a reality. Possessing a high cell viability and desired shape, their gum grafts are bringing personalized bioprinted tissues into focus.
Digital gynecology: AI to create a personalized pessary device
Building robust AI models depends on quality measurements. For Cosm Medical, that means redesigning the way that pelvic anatomy is imaged and measured.
Aye Nyein San, Head of Technology and Operations at Cosm, describes how the company is creating personalized pessaries made from 3D-printed molds to support the pelvic organs for individuals with prolapse. Their rapid AI segmentation models enable them to automatically identify key pelvic structures from ultrasound images, create patient-specific measurements, and design custom pessaries that improve symptoms and increase patient satisfaction.
The company is researching better methods of analyzing pelvic anatomy, such as their work on developing “colpodynamic imaging,” which uses a water-filled bag to distend the vagina for measuring mechanical properties. These measurements can provide improved diagnostic insights by creating a more objective, quantifiable model of patient anatomy while also improving the accuracy of soft tissue segmentation by AI models due to the higher contrast that the water bag provides in ultrasound imaging.
AI infrastructure is so much more than software. Without accurate and cost-effective instrumentation to objectively collect patient measurements, the AI is useless. Additionally, while cloud-based AI solutions may eliminate the need for on-site computing hardware, if the AI software is slow and laggy, confidence in the tool will deteriorate. For Cosm Medical, tackling the AI challenge means closely examining every step of the process that AI will affect and be affected by.
A ChatGPT for 3D printing
The future of large-language models (LLMs) is personalization: getting tailor-made responses from ChatGPT that answer in your domain-specific language and with a high level of technical knowledge. And innovators are asking: how can we make a ChatGPT for 3D printing that people will actually use?
Dr. Gregory Hayes, Senior Vice President of Global Additive Minds at EOS, notes how EOS is leveraging many years of printer data, user manuals, service reports, and other company records to create robust datasets to train AI to enhance their products. Their ChatGPT-like LLM acts as a knowledge management tool, enabling users to quickly receive advice on how to service an EOS printer without having to manually look through lengthy documents.
The additive manufacturing company, which serves a variety of industries outside of healthcare, is also feeding the data from the sensors of their printers into an AI algorithm to recognize when prints have gone wrong and make corrective decisions on the fly.
Dr. Hayes points out that AI is enabling the large-scale integration of many forms of data: images taken during prints, material properties, sensor data, machine logs, and more. However, AI is still in its infancy, and it’s important to recognize all the factors that AI has yet to take into account. Cell viability? Implant rejection? Likelihood of infection?
As Dr. Hayes noted, the hope behind AI is that it will make printing easier: you no longer need to have many years of experience in printing one particular thing to get the results you want. Troubleshooting, anomaly detection, and error correction can be done automatically or with relative ease using an AI assistant.
What we’re thinking
The critical need is for human innovators to guide the way that AI integrates all these different forms of data together. We might not be able to throw everything in one pot, train an LLM on it, and hope it works out. Instead, it will be important for 3D printing experts to pay attention to developments in making AI “think” in a more methodical manner, using complex reasoning skills and having the AI model explain where and how it got to its conclusions.
While there are still many challenges to get to the point where AI in 3D printing truly makes life easier for all, AI developments have and will continue to depend on the one thing that can be counted on: human creativity. And with the many applications of AI we’ve seen from our speakers, the future of 3D printing is bright. Subscribe to 3DHEALS to join us live for our future online and in-person events.
About the Author:
Peter Hsu is an editorial intern for 3DHEALS. He is currently an undergraduate at the University of Illinois Urbana-Champaign and studies bioengineering with a focus on cell and tissue engineering. He is also minoring in computer science with interests in artificial intelligence and image processing. Peter conducts research on using computer vision methods to analyze human tissue images and improving the robustness of machine learning workflows. He is interested in the use of AI to assist tissue engineering and bioprinting research for medical applications. He is passionate about science communication and leads STEM outreach lessons at schools in the central Illinois area.
Related Links:
Expert Corner: AI in Healthcare 3D Printing: The Future is Now
Microfluidic Devices and 3D Printing (On Demand, 2025)
3D Printing Organ on a Chip, Microfluidics Devices (On Demand, 2023)
3D Printing and Microfluidics (On Demand, 2022)
Microfluidics, Technology, Commercialization (On Demand, 2021)
Event Recap: 3D Microfabrication
Event Recap: 3D Bioprinting Biofabricating Skin Components
Revolutionizing 3D-Printed Facial Titanium Implants Polishing
Comments