Developing XAI approaches for medical experts
A warm welcome to Grégoire Montavon, who is taking up a BIFOLD-Charité professorship on April 1, 2025. The professorship is one of the several planned BIFOLD professorships at Charité – Universitätsmedizin Berlin, the institutional partner of BIFOLD. Grégoire Montavon holds a PhD in Machine Learning from the Technische Universität Berlin and a Master's degree in Communication Systems from École Polytechnique Fédérale de Lausanne. For the past several years, Grégoire Montavon has been a guest professor at the Freie Universität Berlin and a research group lead at BIFOLD.
His new research group "Explainable Machine Learning in Medicne" is affiliated with the German Heart Center at Charité (DHZC) and will be located at the Campus Charité Mitte. „My research group will focus on the development of Explainable AI (XAI) approaches that are informative and actionable for medical experts. A particular focus will be on developing approaches that integrate well with state-of-the-art machine learning (ML) models used in medical diagnosis and research. The work will be a mix of basic and applied research“, explains Grégoire Montavon.
The group will continue a successful collaboration with researchers at LMU and the Medical University Essen on the use of ML/XAI to extract markers from large-scale multimodal data in the context of cancer diagnosis and prognosis. Grégoire Montavon is also starting two new BIFOLD agility projects: one with colleagues at the DHZC, which aims to extract relevant clinical variables for predicting cardiac outcomes, and a second in collaboration with the Max Delbrück Center (MDC), which aims to infer large gene regularity networks using structured generative models and XAI. Two additional researchers will join the group in the first few months to support these research efforts.
On the basic research side, the group will continue a long-standing collaboration with the Machine Learning group at TU Berlin and Fraunhofer HHI to address current challenges in XAI. This includes extending XAI to more complex ML models and ensuring that explanations remain informative and reliable, as well as developing more detailed forms of explanation based on interactions between features or abstract concepts with higher explanatory power.
The professorship will also contribute to the development of interdisciplinary teaching. „The courses I plan to develop will introduce the fundamentals of machine learning, along with practical use cases. The latter will highlight the power and versatility of ML models, as well as the value of XAI in uncovering hidden relationships in the data and making ML models more transparent“, explains Grégoire Montavon.