BioMedical Data Science

Through technological advances (e.g. DNA sequencing, new imaging technologies, ...) the life sciences have turned into a data heavy science. There is huge potential to discover novel insights and address big challenges of our society using these data.

Our research focuses on development and application of (bio)informatic methods and tools to gain insights from complex data sources. We apply these methods in two main areas: Molecular Ecology and Bio-Medical Imaging. At first glance, these topics seem to have nothing in common. But in both these fields large amounts of diverse data (images, text, DNA, quantitative measurements, ...) are available, and we are not yet able to take full advantage of this rich pile of information. In both fields, there are two major challenges:

• Every data type only holds a fraction of the available information (like pieces of a puzzle). It is necessary to properly combine all data sources to gain new insights that can be used to recommend actions or improve diagnoses and therapies. Thus, one key research area of our group is multimodal methods.
• In both fields, there is a so-called implementation gap. That means, that new methods (like machine learning and artificial intelligence) have been shown to have impressive performance but are still not used in practice. Reasons for this gap include a lack of explainability and interpretability of how these black-box models work. Further concerns are an apparent lack of generalizability, transparency and fairness. Thus, the focus of our group is on bridging this gap.