Raphael Sznitman heads the Center for Artificial Intelligence in Medicine (CAIM) in Bern, which was founded a year ago. He wants to promote technologies with real added value, he says in an interview, and reveals how patients can already benefit from artificial intelligence today.
In March 2021, just over a year ago, you opened the Center for Artificial Intelligence in Medicine (CAIM) in Bern. How is the center doing today?
We have had a very successful first year and can look back on many important milestones in our research, teaching and network we were able to achieve – even despite the pandemic: We have published a good number of scientific papers on the application of AI tools in neurology, surgery, eye care, nutrition, hearing and cancer care, to name just a few. We have started a new master’s program for AI in medicine for young engineers and we have launched our research fund and selected the first pilot projects our Center will be supporting.
CAIM is a collaboration of several institutions from Bern. What makes the area of Bern stand out when it comes to the development in the medical and AI area?
CAIM capitalizes on the unique constellation in Bern that joins players from science, healthcare and industry who have explored synergetic collaboration for decades. Bern has a long tradition of developing new technology for medicine and has pioneered in embedding biomedical engineers directly into the clinic. The founding of CAIM by the University of Bern’s medical faculty and the Inselspital, Bern University Hospital, is a natural continuation of this philosophy of bringing engineering, technology, and medicine together. And with the University Psychiatry Services (UPD) and the Swiss Institute for Translational and Entrepreneurial Medicine, sitem-insel, as partners we cover the entire medical field and the topics of commercialization and spin-offs.
Other Swiss universities have recently launched their own AI research centers (though not focusing specifically on medicine). Are you planning to cooperate with one of them?
As you are mentioning, CAIM focusses on medical applications of AI because we have a lot of expertise here. Of course, we are happy to reach out to and collaborate with our Swiss colleagues working with similar approaches. But part of the motivation behind founding CAIM is also to tap into the yet unexplored synergies within the Bern Medical Hub itself by connecting experts.
Switzerland as a country has a rather bad reputation when it comes to e-health (or digitalized medicine). As a researcher, do you agree with that?
In the beginning Switzerland was rather slow in the digital transformation of healthcare but the last five years that trend has turned around quite effectively and we are seeing a large number of e-health projects and initiatives. So, I believe that the mid-term outlook is very positive for Switzerland.
How does the state of Switzerland in e-health impact your and CAIM’s work?
Part of the research at my lab and at CAIM is interested in cutting-edge technology. We are constantly exploring new opportunities and how the forefront of technology can impact healthcare systems and patients. The current state of e-health in Switzerland does not directly impact this research itself but only comes to play for its integration into the healthcare system later.
One part of CAIM is a Research Fund, which you mention in another interview. Can you reveal some of the projects supported by this fund?
Sure: This spring we have selected five projects that our Fund will be supporting over the coming two years. The projects come from different medical disciplines and strive to solve some of today`s greatest challenges in acute medical care, the prevention of high prevalence health risks, and the provision of tailored just-in-time care for chronic diseases. They cover heart muscle inflammation, kidney stone prevention, monitoring of multiple sclerosis, support for nurses during night shifts and the development of an app that allows women in menopause to track their personal health risks. All the projects aim for a tangible patient benefit – which is important to us at CAIM, as we do not want to develop technology without real added value.
What have been some of the challenges you faced during the first year of CAIM?
The pandemic has been a challenge for all research institutions and CAIM is not an exception. As an important part of our activities if the strengthening of our network, it was a pity that we could not have an in-presence opening event last year. But we still managed to get an interested audience to join us online. We are now looking forward to a post-pandemic area when physical meetings become feasibly again and, for example, are working on a public appearance at the University of Bern´s Researchers’ Night in September.
If you had unlimited funds, what specific CAIM project would you like to realize?
Our goal is to impact patients’ quality of life with new ideas and therapeutic approaches connected to AI. If we had unlimited funds, we would focus on building strong innovation pipelines. These would include fostering scientific discovery and innovation but also integration into the commercial landscape and the healthcare system to effectively make a difference in medicine.
On your personal website, you write your research interests are in the domains of medical image analysis, computer vision and machine learning. What originally inspired you to start researching those areas?
What attracted me to AI was how technology mimics human-like behaviors. It seems intelligent but isn’t really in the same way that humans are, as it only works for very specific tasks. I started out studying neuropsychology to understand the learning process and later transitioned into a more technical area, wanting to know how machines learn and process new information. Then, I was strongly interested in bringing technology to the patient or the physician to support new and personalized treatment approaches.
You've been studying artificial intelligence, computer vision and so on for many years now. What is something you know today you wished you would have known right from the start?
A very important point is to always stay open-minded. When I first started working with clinical experts, I realized that their “language” and their way of approaching a medical problem is very different from an engineer’s approach. If you dare to look outside your own domain, you can connect the two worlds and reach a deeper understanding of how an AI technology needs to be done. Once you have a team of people with that shared goal, then projects become very powerful. And I believe this mindset and a continuous drive to iterate, improve and validate your own ideas are at the core of all successful AI projects in medicine.
You also lead a research group on AI in Medical Imaging with a focus on Eye Care. When I hear computer vision and artificial intelligence, I spontaneously think of you developing some kind of artificial eyes. Are you?
Well, so far, we have focused more on developing AI tools that can help ophthalmologists assess imaging of the eye in a faster way by identifying important biomarkers in eye scans. In the future, it is conceivable that the eye with its wealth of information could serve as a general “window” to the body insofar that it gives indications vascular or neurological diseases which an AI can identify and evaluate.
What are some of the practical uses for ophthalmology developed?
Artificial technologies can, for example, analyze an extensive data set of eye scans in just a few seconds and mark areas with inflammation or fluid deposits, indicating a variety of chronic eye conditions. Because these tools are fast and accurate, they allow eye specialists to assess during parent consultation, how quickly a chronic disease such as AMD is progressing and how often patients need to receive treatment and come for follow-ups. To date, this has been a tedious manual task for doctors, who need to see an ever-growing number of patients with chronic eye diseases every year. Here, AI not only helps from an organizational point of view (planning consultation capacities), but also gives patients improved security that they will not miss key moments to delay disease progression.
We have been talking a lot about research, about emerging projects in artificial intelligence. How much can clients (visiting the hospital) profit from this research today?
The technology can essentially help patients to receive a more tailored treatment to their individual needs and situation. Eye care is one area where AI systems are beginning to help here as outlined above. In radiology, we also have AI implemented in clinical routines; it has, for example, helped assess covid-19 infections during the pandemic, distinguishing them from other respiratory tract diseases in chest x-rays. An area where AI is already very strong a healthcare apps, such as apps for dietary monitoring or special diet needs (diabetes etc.). But AI is also emerging in surgery planning and execution (making interventions even safter) and in cancer care (distinguishing tumor types to find the right therapy or adapting radiation dose to protect healthy tissue).
Coming back to CAIM, what are the next steps with the center?
We look forward to the second year of our new master’s program for AI in medicine with the first thesis projects and to the students that will be starting the program in fall. We also plan to have a symposium on our research in early winter and to intensify the work of our Embedded Ethics Lab by continuing a successfully pilot of Ethics Talks and establishing trends on the social acceptance of AI in healthcare.
