Type 2 diabetes generates high healthcare costs in Switzerland and causes significant suffering for those affected. A multidisciplinary Innosuisse Flagship project led by Empa aims to use artificial intelligence (AI), innovative digital sensors, and new business models to facilitate the prevention of this widespread disease.

Type 2 diabetes is a common disease of affluence. Around half a million people in Switzerland are affected by this chronic metabolic disorder - and the trend is rising. In type 2 diabetes, the body's response to the hormone insulin is impaired. Blood sugar levels rise - and with them the risk of numerous complications, including heart attacks and strokes.

Although genetic factors also play a role, it is primarily our lifestyle that makes this disease a widespread public health issue. Lack of exercise and an unbalanced diet lead to obesity and can also contribute to the onset of diabetes over the years. The good news: A change in lifestyle habits could delay or prevent a large proportion of type 2 diabetes cases. The bad news: For most people, such a change is anything but easy.

So how can we make diabetes prevention more effective and accessible - not only for patients, but also for food companies, health insurance providers, and the healthcare system? This is the question addressed by the multidisciplinary Innosuisse Flagship project PreDiabeTx, led by Empa (see box). The project consortium consists of seven research institutions and eleven implementation partners, ranging from health insurance companies and hospitals to the public sector and the food industry. Together, the partners aim to develop and implement digital technologies for the prevention of type 2 diabetes in Switzerland.

A holistic approach

PreDiabeTx addresses three major challenges in prevention. The first is the low long-term participation in preventive measures. The project partners emphasize that this is not merely a matter of individual behavior. Systemic factors also play a significant role here, such as the structure of incentives, the accessibility of measures for different segments of the population, and funding. The project's other two focal points are therefore the lack of sustainable business models for prevention among healthcare providers and health insurance companies, as well as the still insufficient use of health data to enable personalized support.

"Diabetes generates high healthcare costs in Switzerland," explains Empa researcher Thijs Defraeye, who is leading the project.

By collaborating with partners from various fields, we hope to develop holistic prevention measures that will actually be implemented. Prevention is difficult to measure and quantify. If health insurance companies and businesses cannot quantify the cost-effectiveness, they find it difficult to invest in prevention measures."

Simon Annaheim, Empa researcher 

This is precisely why one of the five subprojects of PreDiabeTx, led by the University of St. Gallen, focuses on sustainable financing and incentive models. "Effective digital diabetes prevention alone is not enough; it must also fit into the healthcare system and be mutually beneficial and sustainable. In our subproject, we are therefore working with CSS, Helsana, Sanitas, and SWICA, as well as cantonal authorities, to develop concrete business and incentive models that can establish digital therapies as an integral part of standard care," says Elgar Fleisch, professor at the University of St. Gallen and head of the subproject.

Eating healthier thanks to AI

Two other subprojects focus on digital tools for diabetes prevention that can use AI to provide personalized advice for a healthier lifestyle. "We want to develop a 'digital coach' that supports people while shopping and cooking," says Empa researcher and project leader Cynthia Sob. According to the researcher, one's own health can be significantly influenced by diet.

This so-called "food as medicine" approach is gaining increasing importance in medical research - but is not yet widespread in Switzerland. "In the U.S., there are already health insurance companies that issue prescriptions for healthy foods to promote healthy eating. We want to investigate to what extent comparable approaches would also be possible within the Swiss insurance system," explains Defraeye. This would make vegetables a bit more financially appealing than chocolate croissants.

This also benefits people with lower socioeconomic status - who are already at higher risk of developing type 2 diabetes. A separate subproject of PreDiabeTx is dedicated to them and other at-risk population groups. As part of this project, the partners are developing digital coaches tailored to their needs, which take specific requirements into account and provide personalized, easy-to-follow advice.

Sensor T-shirt for monitoring success

The last two subprojects rely on non-invasive, everyday-use measurements to demonstrate the success of the preventive measures. One of them, led by the University of St. Gallen, focuses on wearables - devices such as smartwatches that continuously monitor vital signs. The second, led by Simon Annaheim at Empa, relies on textile sensors. "We are convinced that occasional measurements at the doctor's office or the pharmacy do not provide a sufficiently accurate picture of one's health," says Annaheim. The researcher and his team have therefore specialized in developing soft textile sensors that can be integrated into items such as a chest strap or a T-shirt, allowing vital signs to be recorded comfortably and discreetly in everyday life. Unlike wrist-based measurements, as performed by most smartwatches, the textile sensors on the torso are somewhat more complex. In return, they provide more accurate readings. According to Annaheim, they could thus serve to monitor health status over several days.

What all the measurements in the project have in common is that they are non-invasive - that is, they do not penetrate the skin. While blood sugar cannot yet be reliably measured this way, the other readings - such as heart rate, respiratory rate, blood pressure, and oxygen saturation - do allow for conclusions about one's health status. The researchers aim to better understand how the measured vital signs relate to diabetes risk using AI-powered algorithms.

Helping instead of lecturing

This is what intelligent and sustainable Type 2 diabetes prevention could look like in the future: An AI coach provides personalized dietary advice tailored precisely to the user's specific situation, making it easier to follow. Targeted incentive systems and programs offered by health insurance companies and other organizations further facilitate a healthy lifestyle. And through wearable sensors, doctors and patients themselves monitor progress.

"We don't want to make anyone feel guilty or scared but rather offer support and make a healthy lifestyle more accessible," says Simon Annaheim. The benefits are far-reaching: "By preventing type 2 diabetes, you also prevent other health problems, such as certain types of cancer," says Cynthia Sob. "But unlike cancer risk, the reduction in diabetes risk can be measured much more effectively in a short period of time." The researchers are convinced that when patients quickly see and feel their health improving, they gain additional motivation.

There is still a lot of work to be done before this vision for the future is realized. PreDiabeTx launched at the end of 2025 and will run through the end of 2029. "Of course, we won't solve a problem that has developed over decades in just four years," says Annaheim. "But we hope that, through our complementary and holistic approach, we can take diabetes prevention a step further together with our implementation partners."