Dr.-Ing. Mario A. Cypko

ME AS A SCIENTIST

​As medicine becomes increasingly digital and AI-supported, the key challenge is not only to build intelligent systems, but to keep them understandable, controllable, and useful for clinicians and patients. I am interested in medical technologies that improve patient safety, decision making, and quality of life while preserving meaningful human oversight.

​

In my research group Medical AI and Modelling (MedAIM) at Hahn-Schickard, we focus on knowledge modelling for cancer care and transplant follow-up. Another major focus is closing the gaps between clinical snapshots through the continuous capture of patient vital signs and behaviour using smart wearables. More broadly, my work explores trustworthy clinical AI, multimodal patient representations, and Model-Guided Medicine. I am particularly interested in connecting data-driven AI with explicit, inspectable clinical models and in keeping the human-in-the-loop wherever transparency, responsibility, and meaningful clinical control matter.
 

A growing part of my work is the idea of Medical Model Science, first described by my doctoral supervisor Prof. Heinz Lemke, which I advocate as a necessary complement to Data Science and Medical Informatics. My interest here is in treating clinical models not merely as technical by-products, but as scientific and operational objects in their own right: models that need to be understandable, reviewable, maintainable, governable, and linked to clinical purpose and context. This includes questions of transparency, validation, lifecycle management, human-AI interaction, and the responsible integration of AI into real-world medicine.
 

My background is in Medical Informatics, with work on AI-based clinical decision support, multidisciplinary decision making, Bayesian and knowledge-based modelling, and human interaction with computational medical models. During my PhD at the Innovation Center Computer Assisted Surgery of Universität Leipzig, I worked on complex cancer cases and model-based decision support. I then continued as a PostDoc at the European Space Agency in the Netherlands and later at the German Heart Center in Berlin, where I explored both the capabilities and the limits of AI in demanding medical settings. Through the Excellence Cluster Matters of Activity, at the Human-Centered Computing group at Freie Universität Berlin and with the Charité Berlin, this perspective expanded further towards the question of how much AI we actually want, where we need it, and where human-centred design and the human-in-the-loop remain essential.​