"The human being is always at the centre of our research"
The second interview in our portrait series was conducted with Klaas Enno Stephan, Deputy Head of Department at D-ITET and Professor at the Institute for Biomedical Engineering (IBT) of ETH and the University of Zurich. He talks about his cooperative working environment, explains how mathematical models might help to find the right therapies for psychiatric diseases and how his research contributes to building bridges between medicine and engineering.
Prof. Stephan, what is your research about?
I do research in two still relatively unknown areas that are closely linked. Translational Neuromodeling develops and validates mathematical models based on concrete concepts of brain diseases. These models are then used in an attempt to infer hidden processes responsible for the diseases from behaviour, physiology and brain activity. In the second, adjacent field, Computational Psychiatry, the computer-aided models are tested in clinical studies.
In other words, in a first step, we develop mathematical methods that are subsequently implemented in computer models. We then conduct patient studies in our own clinical unit.
How does your research approach differ from current methods of machine learning?
We focus on the human being, driven by theories about what causes mental disorders. That is, our models are mostly formulated from a medical point of view for a very specific disorder in the brain. They are a reflection of how a certain psychiatric disease is conceptualized by medicine today. Thus, our models are intended to provide concrete predictions and understanding: not only why a disease develops, but also what it means for the patient.
This approach thus differs greatly from classical methods of machine learning. In machine learning, for example, all available brain data would be fed into a neural network and a prediction (such as a prognosis) would be obtained. It is difficult, however, to interpret this prediction in mechanistic terms. In our approach, however, we can apply a very specific model to the individual and estimate physiological or cognitive processes that we can interpret. In a further step, however, we do use machine learning methods as a tool to create predictions from the parameter estimates the model provides.
What kind of studies are you currently conducting and how are they financed?
We are interested in physiological processes that can be influenced by psychotropic drugs as well as cognitive processes, treated with psychotherapy. However, we deliberately do not accept any support from the pharmaceutical industry, we are only financed by public funds and charities. Currently we conduct six patient studies on schizophrenia, depression, autism, multiple sclerosis, gambling addiction, and pain.
All studies deal with clinically relevant questions, such as "can we predict for an individual whether a certain therapeutic approach, be it a drug or a psychotherapy, will work?" or "can a patient who has benefited from medication and who is now well again be advised to stop taking it?”.
On the other hand, we also carry out basic science studies with volunteers without psychiatric diagnosis in order to establish methods as a basis for later patient studies.
You studied medicine, computer science and mathematics at the same time. How did this come about?
To avoid misunderstandings: Of these three subjects, I have only completed medicine. Unfortunately, there was not enough time for my diploma thesis in computer science, as I immersed myself in research right after completing my medical studies. Later, however, I completed a doctorate in neuroinformatics, so that I was finally able to bring the two disciplines together.
“There are two hearts beating in my chest: one for complex systems and their intellectual challenges, and one for medicine that directly helps people.”Prof. Klaas Enno Stephan
I was therefore not completely unsuitable for my current chair at the interface between medicine, computer science and mathematics. I could have explored the immune system, or the heart, but the brain is just amazingly fascinating, which was the deciding factor. In addition, I saw the greatest clinical need at that time in psychiatry, which is still the case today.
Your chair, the Translational Neuromodeling Unit (TNU), was established in 2012 and is part of the Institute for Biomedical Engineering (IBT) at ETH and the University of Zurich (UZH). How well do you feel integrated into D-ITET?
Very well! ETH Zurich is a top institution with an enormous breadth and depth. In each area there are fantastic colleagues who work at the highest level of excellence. And the spatial distances between the institutes are small, especially here in the centre. I am also fortunate to be located at a very cooperative institute, the IBT, which is quite important in everyday life. At D-ITET I personally like the fact that a lot of young professors have joined us in recent years. I also appreciate the thematic diversity very much. Furthermore, I would like to highlight the partnership with the University of Zurich, to which I also belong. It is only through the cooperation between ETH and UZH that many projects in the medical field become possible.
How heterogeneous is your group? Are you looking for doctoral students?
My group currently consists of about 36 people from 16 nations and has reached its maximum size. At the moment, we have a slight female majority and we are very interdisciplinary: from medicine and psychology to mathematics, physics and electrical engineering, my team covers a wide range of research expertise. The direct collaboration between medical professionals on one side and computational and mathematical scientists on the other side builds bridges not only between the various disciplines and concepts, but also between UZH and ETH Zurich. In my group, I can sense that everyone has a lot of respect for other disciplines, which I am very pleased about and which I also attach great importance to.
What are currently the most challenging issues in your research field?
One of the greatest challenges for the next decade will be to bridge the gap between brain and body. Psychiatry is very much fixated on the brain, and it has somewhat neglected how strong the interactions between body and brain are. Making this brain-body interaction measurable and quantifying it in models is complicated but clinically very important.
The practical challenge concerns the organisation of the patient studies. Depending on the study and the disease pattern, it can sometimes be very difficult to find enough volunteers. Most of the time we have to search through partner clinics in and around Zurich, and this is the biggest bottleneck: patients only come to us for research purposes, we do not offer them medical care. In addition, the structures in clinics often change rapidly, so that our contacts change frequently. Recruiting patients for studies is not just a problem for us, it is a global problem. Nevertheless, ETH Zurich and UZH have chosen the right strategy by founding the TNU. Through long-term financing, our own clinical staff for study purposes and a patient-friendly infrastructure, ETH and UZH have made it possible that reliable studies can be conducted in the first place. We are very grateful for this.
Professors at D-ITET
In our interview series, professors at D-ITET give an insight into their research and personal motivation to go into academia.