top of page

Dr. med. Jan Clusmann

Medical Doctor & Clinician Scientist

Data Science Track Participant
TechLabs Aachen

June 16th, 2024
 

Experience Report: From a Beginner Course to the Training of Cancer Detection Models

Hi Jan! How has the TechLabs Data Science track influenced your career?

 

"TechLabs kicked off a new direction in my career. In my medical studies, I was initially not really involved in data science or AI. But when I did a lot of manual image and data analyses for my doctoral thesis, I realized that it took far too long and was inefficient, repetitive work. That's when I got ambitious and wanted to dive deeper into the automatization of such processes. I became aware of TechLabs, took the Data Science course in 2021 and I was really enthusiastic about it. After graduating in 2022, I started working as a clinician. But I knew that I also wanted to continue with research, this time focusing on programming and analysis of big data. Individuals with such skills are currently in high demand - not only because they understand coding itself, but also because they can collaborate with many other technical disciplines. Today, I am working (in parallel to my clinical duties) on research projects in big data and computational oncology. We analyze massive datasets with health and lifestyle data from >500,000 individuals and try to derive new diagnostic, screening or treatment methods. I have two main goals right now: the first is to improve early detection of liver cancer. I am working on the development of machine learning models that would detect persons that are at higher risk of developing liver cancer, based on their health records, lifestyle, blood measurements and genetics. The idea is, that these patients could then undergo more frequent screenings. While these models are far from being used in the clinic, the prediction itself is actually going really well, mostly because these models can combine vast amounts of data that humans cannot analyze for every single patient. The second goal is the implementation of large language models (such as the popular ChatGPT) into medicine. The potential for these new technologies is gigantic, and there is a huge interest on models like these in the (medical) scientific community. For me, this actually ment a career change: I recently moved to Dresden, to pursue my research goals full-time as a Postdoctorial researcher at the first-ever German "Institute for Clinical Artificial Intelligence". Looking back, I eventually got into this field through TechLabs and I am really glad I did."

You wanted to start programming for a long time. What were your blockers and how did TechLabs help you overcome them?

 

"Many of my friends studied computer science related subjects, and were struggling with the heavy mathematics that are often involved. It got kind of self-evident for me that one would need these mathematical basics to be able to start programming in the first place. Then TechLabs made me realize that you can start easily without any prior knowledge and that it is mainly learning through trial and error. To be honest, coding is also a lot of googling and searching for other people’s solutions to similar problems as yours. You don't have to reinvent the wheel."

How did you experience the track without any prior knowledge? What’s your key take away?

"I was surprised that coding was much less abstract than I thought and how intuitive it can become.. The theoretical lessons and online-tasks deliver a very soft start. Setting up your programming environment and Github might actually be the hardest task in the first weeks. And once you start working together on the projects, the real fun begins. We actually worked on a project idea that I had proposed myself: Building an object detection model to efficiently evaluate tumor cell clusters, basically automatizing the task I had performed manually for my doctoral thesis. Funnily enough, I did not write a single line of code for the project myself. It was much more relevant for me to discuss the biological/medical concepts with the other, much more technically oriented people in the group. It was amazing to see how the combination of individual strengths of this diverse group unleashed an enormous potential, where each one of us individually would have failed, because of a lack of understanding of either the biological/medical or the technical side of the project. My key take away is, that e.g. as a clinician, you do not have to be an expert level programmer to start working on these subjects. However, a basic understanding of the technical concepts involved in the project (in our case for example object detection models) is essential to even be able to convey an idea. That's where I find TechLabs very strong, and it simulated surprisingly well how working together in an interdisciplinary research group looks like nowadays."

How important do you think it is that more medical professionals acquire data science skills?

"I think it is both essential and in the long-term, inevitable. The potential is just enormous.The medical field is undergoing a major transformation as we generate more and more data which is full of potential but just not possible to grasp manually for clinicians. I experience that many medical students want to develop in this direction, but there is a lack of opportunities to actually acquire these skills. Usually, there is no suitable offer in the curriculum of medical studies, something that should urgently be implemented… I remember that statistics used to be a rather small (and often, a little boring) subject at school. In med school, statistics was a somewhat minor and peripheral subject. I found it rather dry and somewhat unengaging, which unfortunately led me (among many others) to just scrape through with the bare minimum of effort. The vast potential of the subject, especially in today's context with the advent of artificial intelligence, was not sufficiently emphasized or explored. This leaves a lot of potential untapped. I am now trying my best to advertise for courses like TechLabs, which could fill this gap rather quickly. I think that especially to improve the aspect of communication with other disciplines (as I mentioned earlier) TechLabs is a great way to learn about Data Science and AI, especially when you are at an early stage of your studies, ideally before starting with a doctoral thesis. Sadly, many people are hesitant to delve into this topic, fearing the challenges it might pose and perceiving seemingly insurmountable barriers. However, getting started today is easier than ever, especially with the help of tools like ChatGPT and other programming assistants. When I am coding today, I always have ChatGPT open, it can easily work as a personal programming tutor, available 24/7. So, with a combination of TechLabs courses and ChatGPT, there really is no more excuse not to get started."

Thanks for sharing your experience!

bottom of page