Artificial intelligence proves itself in medical technology

When 25-year-old service engineer Claudia Esteban Patón opens up her laptop in the technician room of the Vila-real Renal Care Center in the north of the Spanish city of València, she does so carefully: her laptop has B. Braun’s own customized AI servicing program, Service Assistant. The program displays every dialysis machine in the center. At a glance, Esteban Patón sees that the machines are all working correctly—except for one. The software reports: “critical status.”  The machine has already been taken out of service by center staff. With a single click, Esteban Patón learns the likely cause of the issue: there is a 90 percent chance that the problem is with the balancing chamber diaphragm, one of the machine's core components.

The program shows Esteban Patón the steps needed to make the repairs. In a chat window, she types: “what is the part number for the diaphragm?” The information immediately appears on the screen. Esteban Patón sees that the part is available, smiles contentedly and says, “the Service Assistant helps me with my day-to-day work and lets me concentrate on more important things.” 

Major increase in efficiency predicted

There are few topics currently occupying the health care industry as much as the use of artificial intelligence. For example, AI makes it possible to detect disease early, understand it better and develop just the right medication. The McKinsey Global Institute believes that the pharmaceutical and medical technology industries can use AI to generate additional value of 60 to 110 billion dollars per year. ^[1]

B. Braun is part of this development. “For us, it's about identifying concrete use cases where artificial intelligence really adds value and increases efficiency and quality,” said Dr. Michelle Heber. She is the head of the AI Hub, a team at B. Braun that acts as a sparring partner for employees who are advancing AI projects. Wiebke Zimmermann, an AI consultant at the AI Hub, said, “We want to combine the expertise at the company with the potential of artificial intelligence.”

A machine that helps 

The Service Assistant is a particularly good example of B. Braun's approach when implementing AI. The project began in fall 2023, according to Dr. Philipp Leise, who oversees the development from the technical side. “We worked in an interdisciplinary team from software development, AI development and Technical Services.” Joshua Kirsch, a data scientist at the AI Hub, added. “We had the first prototype within a few weeks.” Alexander Kammenhuber, the head of Technical Services at B. Braun, explained. “In the full sense of Sharing Expertise, we quickly saw success in the project by including different departments. The development of the Service Assistant is setting standards for technical services. Quick and agile implementation of new technology in the service process allows us to keep refining the high efficiency of our services.”

The Service Assistant consists of two parts: First, the program uses AI to analyze sensor data from the machine to determine the exact status of the components. Second, the program offers a chatbot that is based on a large language model, something similar to ChatGPT. This allows the program to interpret the questions the service engineer asks. It scans current service documents for the answer and displays it on the screen.

“Of course, I can look up the part numbers for replacements or the repair steps myself,” said Claudia Esteban Patón. “But with the Service Assistant, it goes much faster and it works perfectly with my daily routine.” She puts plastic gloves and safety glasses on, then begins swapping out the components. She quickly removes the defective diaphragm and holds it up to the light. It has a tiny tear—the AI’s prediction was correct. Esteban Patón explains, “Without the Service Assistant, I would have had to rule out a whole series of different root causes before I would have gotten to the defective diaphragm.”

Digital Doppelganger

Dr. Leise, the project lead, is pleased with the launch of the Service Assistant. “Now it’s a question of working together with the users to expand the app’s functionality so it helps them as much as possible.” In the future, the application will also pave the way for the jump to predictive maintenance; components will be continuously monitored and replaced based on specific usage and wear margins. It is also possible to construct a digital twin of the dialysis machine. This is a virtual representation that can be used for monitoring, simulation or upgrading.

At B. Braun, there are currently a number of AI projects at different stages of completion, such as AI for visually inspecting ampules—called Mini-Plascos®—during manufacturing. Now, specially trained employees need to manually inspect the Mini-Plascos® for damage or the slightest sign of contamination. In the near future, a machine will help perform this task and AI will continuously refine its ability to identify defective ampules. 

Process Optimization

Another domain of AI at the company is translating website content into simplified language. It starts by defining a set of rules, such as reformulating long sentences into short sentences or finding synonyms for foreign words. The AI then takes over, converting the complicated text into accessible language. This is also relevant from a medical view since patients also use the B. Braun website to research their treatment. Clear information provides better understanding of the treatment and supports a successful treatment outcome.

“The health care industry is a particularly rewarding field for the use of AI methods,” said Elsa A. Kirchner, professor and head of the Institute of Medical Technology Systems at the University of Duisburg-Essen, Germany. “AI helps improve and sometimes even redefine already advanced technology and methods, whether in pharmaceutical development, robotics and prosthetics, everyday hospital operations or the production and maintenance of medical devices.” She explains that, given the considerable amounts of money our national economies invest in health care, AI could offer much needed increases in efficiency. It could also make it possible to give patients and providers better, more customized support. Kirchner has high hopes for the European industry. “We are world-class in research. When it comes to application, though, we are still just getting started despite all of our success so far. We need to catch up quickly. In Europe, we can demonstrate what using AI looks like when we focus on the benefits for people.”

Humans are still accountable

Of course, the concept of accountability is also central to B. Braun. “Data security and protection are top priority in all our projects,” said Dr. Heber at the AI Hub. “We need to always make it clear how and where the AI gets its information.” And something else is important to her: “We don't want to replace human labor, we want to augment it.”

In Vila-real, Esteban Patón has replaced the defective diaphragm and is installing the balancing chamber back into the dialysis machine. “When I was a kid, I took apart and fixed everything I could, from pens to radios.” Her favorite tool is a cordless drill with a flexible attachment that can get too hard-to-reach fasteners. “I also think of AI as a tool, it's a really fascinating one.” With all our amazement at what AI can do, we should never rely on it entirely, added Esteban Patón. “I still need to check whether the AI's prediction is correct myself , whether makes sense with what I’m seeing, what my experience tells me.”

Once the repair is complete, the Service Assistant shows her the tests that still need to be performed to check whether the dialysis machine is fully functional again. Everything runs smoothly. Esteban Patón gathers up her tools and closes her laptop, explaining, “you get the best results when humans and machines work together.”