How Can AI Make Medical Device Manufacturing Better?
Artificial intelligence (AI) is improving outcomes in a wide range of sectors. It can help companies increase output while keeping costs down. This article looks at the specific benefits that AI could bring if used as a medical device manufacturing technology.
Making AI Medical Devices Smaller and More Powerful
A major downside of deploying AI on a medical device now is that the algorithms require significant computing power. That could mean AI-based medical devices are not as small as they need to be to make them maximally user-friendly and versatile for providers and patients.
However, some researchers are working in the area of TinyML. They want to figure out how to make AI microprocessors small enough that they can run locally on medical devices. One of the people specializing in this work is Song Han, assistant professor in electrical engineering and computer science at MIT.
In an article published in Medical Device Developments, Han explained, “We have full-stack innovations from efficient algorithms to systems and hardware. “When the hardware device shrinks down, the power, computation and memory budgets become very tight and make it difficult to run advanced AI algorithms.”
He believes TinyML can accommodate those challenges. That’s why he’s simultaneously designing hardware and algorithms that could improve future medical devices. A recent project involved applying AI to take blood pressure readings.
“Many workloads in the medical sector can benefit from AI algorithms,” he said in the article. “The efficient deployment of AI algorithms on small devices will empower smart medical devices.” He hopes his work will make smart medical devices more “hardware-aware,” so the algorithm would automatically recognize a device’s hardware parameters and adjust accordingly.
Progress in this area would help medical device companies find new opportunities for product utilization. AI could improve device manufacturing in other ways, too, such as by highlighting potential design improvements. Some of today’s 3D printers have AI capabilities that help manufacturers reduce waste or pursue additional optimization. 3-D printers can also accelerate prototype production.
Revealing Trends About Medical Device Failures
Even the most carefully designed and thoroughly tested medical devices sometimes have unintended consequences once patients begin using them. However, a recent project used AI to uncover some gender-based trends about those failures.
When the researchers initially used the FDA Manufacturer and User Facility Device Experience (MAUDE) database, they ran into a problem. That public dataset contains 8 million reports about incidents when medical devices malfunctioned or otherwise put people at risk. So, it seemed ideal, except that MAUDE does not include gender-differentiated data, and researchers knew medical devices can affect men and women differently.
Fortunately, a closer look at the available data showed that the descriptions in those reports often included pronouns. After realizing that, the researchers built an AI algorithm that correctly identified the sex in more than 340,000 medical device injury and death reports. Additionally, it showed that 67% of the patients were women, and 33% were men.
The organization that carried out the research is now pushing the FDA to do more to keep people safer from medical device dangers, including releasing the full dataset to the public about recorded incidents. Then, patients, providers and device makers would have more details about what went wrong and why. They could use that information to shape their decision-making, and in the case of manufacturers, make changes that could prevent future issues.
Such AI applications could also cut excessive costs at hospitals. Statistics indicate supply chain costs account for approximately one-third of U.S. hospitals’ operating costs. Looking more closely at what makes medical devices fail could reveal that certain materials break down in the body faster than others. Addressing that problem could reduce the surgeries and implanted device replacements that many patients must undergo to prevent complications.
Accelerating Inspection Speeds While Raising Accuracy
Computer vision is a subset of AI that often gets applied to quality control and inspection tasks. Since medical devices must satisfy such tight regulations, this type of medical device manufacturing technology could support a manufacturer’s bottom line and reputation.
People familiar with the industry say that the need for such precise quality control within the medical device sector could result in hundred-fold price increases for parts that were initially fairly inexpensive. Tom Brennan, president of Artemis Vision, told Vision Spectra, “You’re inspecting quality into the part, and you’re increasing the price because it’s to a guaranteed quality.”
Some of the most robust computer vision systems can detect pits or bumps that are only the width of a human hair. It’s also necessary to determine whether such defects are cosmetic or if they could negatively impact how well the device works months or years down the line.
One of the primary benefits of such AI applications is that they can get consistently reliable results without experiencing the fatigue that humans do. Conscientious employees still get tired eventually, so their ability to spot problems with medical devices may gradually decrease as their time on shift lengthens.
Computer vision can also verify that a medical device component looks the same on both sides, when applicable. Alternatively, it can check that the devices have the necessary identification numbers or marks. Thanks to the versatility of computer vision, manufacturers can choose the most effective ways to apply it based on existing bottlenecks or quality control goals.
Raising Productivity in a Challenging Marketplace
Artificial intelligence can also be valuable to deploy when manufacturing decision-makers realize they must make changes to remain competitive against other organizations. Leaders at Argon Medical Devices were in that position when they chose to deploy collaborative robots in their factory several years ago.
At the company’s 85,000-square-foot facility, production line workers assemble more than 4 million needles and catheters per month. Tim Lenehan, director of plant operations, told Assembly Magazine, “We invested in automation because, like any manufacturing facility, we are continually looking to reduce our costs. In addition, we are looking to relieve our labor force of redundant operations so they can focus more on value-added tasks.”
The automation equipment used in today’s manufacturing facilities relies on artificial intelligence in several ways. Some machines need it to perceive their surroundings, ensuring they can work safely around humans. They also use AI to learn tasks once humans teach them. Many industrial robots even show improved performance through extended use.
Kevin Hess, senior engineering manager at Argon Medical Devices, explained how collaborative robots have been game-changers for the company. “Historically, we have focused on hard automation to automate our floor. However, with the advent of collaborative robots, we have been able to automate areas which were previously undesirable for hard automation.” He also specified that automation has tripled throughput and cut scrap rates by about 10%. Additionally, direct labor costs have decreased by approximately 30%. That change helps the company maintain competitive prices without sacrificing quality. Using automation also helped the company raise its capacity by 7.5%, supporting the facility’s future growth.
AI Is a Vital Medical Device Manufacturing Technology
These examples highlight why it’s increasingly common for medical device manufacturers to at least consider using AI to support production. Figuring out the most appropriate ways to implement it takes time and money, but the results are often worth the effort.
Related Articles
-
There are several tactical approaches to design that can help medtech manufacturers inspire patient confidence in their products and ensure proper usage. Following are three ways you can help ease the burden of delivering and receiving quality care through product…
-
At Device Talks Boston in May, Ronald Kurz, Sr. Director and GM at Canon Virginia, Kathryn Unger, Sr. Mgr. of Global ESG Communications at Boston Scientific, and David Ettl, COO at Gradian Health System, addressed the question, how can we…
-
“This report validates the broad applicability—and true value—of computational modeling and simulation as a critical engineering discipline. Tangible cost, quality and performance improvements are being realized by the early adopters.”
-
As technology component pricing continues to rise, MedTech manufacturers are applying proven methods and exploring new approaches to maintain profitability and extend product lifecycles.
