Jun.-Prof. Dr. Maria Kalweit

Chief Scientific Officer (CSO)

With a background in artificial intelligence, Dr. Maria Kalweit, CSO of CRIION, focuses on the development of end-to-end systems that move toward real-world medical application. Her work spans the entire translational pipeline — from automated detection of cell types and states in microscopy images to the development of adaptive treatment strategies that continuously refine themselves over time.

While grounded in AI methodology, her current focus lies on bridging model development with experimental and clinical translation. This includes the deployment of robust and interpretable systems in settings characterized by limited data, evolving input distributions, and clinical complexity. Her work integrates technical innovation with software pipelines, interactive tooling, and experimental validation — always with the goal of making AI meaningful and actionable in oncology.

Working on challenges of bringing AI to the real world:

  • Efficiency: satisfy runtime, data or memory constraints
  • Interpretability: explanations for the models decisions
  • Accessibility: make AI available for non-experts
  • Trustworthiness: reliable, fair, transparent, and safe for all users
  • Deployment: provide usable interfaces to AI models

Publications

2025

  • Gabriel Kalweit, Anusha Klett, Paula Silvestrini, Jens Rahnfeld, Mehdi Naouar, Yannick Vogt, Diana Infante, Rebecca Berger, Jesús Duque-Afonso, Tanja Nicole Hartmann, Marie Follo, Elitsa Bodurova-Spassova, Michael Lübbert, Roland Mertelsmann, Joschka Boedecker, Evelyn Ullrich, and Maria Kalweit. Leveraging a Foundation Model Zoo for Cell Similarity Search in Oncological Microscopy across Devices. Frontiers in Oncology. Accepted, appears 2025. web
  • Yannick Vogt, Maria Kalweit, Maria Alieva, Evely Ullrich, Joschka Boedecker, and Gabriel Kalweit. AI in modular concepts of natural killer cell therapy. In Natural Killer Cells, J. Zimmer and E. Ullrich. Springer, Berlin, Heidelberg. Accepted, to appear in 2025.
  • Gabriel Kalweit, Evelyn Ullrich, Joschka Boedecker, Roland Mertelsmann, Maria Kalweit. AI in Optimized Cancer Treatment: Laying the Groundwork for Interdisciplinary Progress. Oxford Open Immunology, Issue “AI-Driven Immunology: Accelerating Discovery and Innovation”. web
  • Mehdi Naouar, Hanne Raum, Jens Rahnfeld, Yannick Vogt, Joschka Boedecker, Gabriel Kalweit and Maria Kalweit. Salvage: Shapley-distribution Approximation Learning Via Attribution Guided Exploration for Explainable Image Classification. The Thirteenth International Conference on Learning Representations (ICLR), 2025. web
  • Jens Rahnfeld, Mehdi Naouar, Gabriel Kalweit, Joschka Boedecker, Estelle Dubruc and Maria Kalweit. A Comparative Study of Explainability Methods for Whole Slide Classification of Lymph Node Metastases using Vision Transformers. PLOS Digital Health, 2025. web
  • Gabriel Kalweit, Maria Kalweit, Ignacio Mastroleo, Joschka Bödecker und Roland Mertelsmann. Künstliche Intelligenz in der Krebstherapie. Künstliche Intelligenz in Forschung, Lehre und Hochschule, edited by Manfred Löwisch, Thomas Würtenberger, Max-Emanuel Geis and Dirk Heckmann, Duncker & Humblot, 2025, pp. 425-434.
  • Maria Kalweit und Gabriel Kalweit. Warum wir neu lernen müssen, mit Maschinen zu sprechen – eine Momentaufnahme der Generativen Künstlichen Intelligenz im Januar 2024. Künstliche Intelligenz in Forschung, Lehre und Hochschule, edited by Manfred Löwisch, Thomas Würtenberger, Max-Emanuel Geis and Dirk Heckmann, Duncker & Humblot, 2025, pp. 111-127.
  • Marieke Wesselkamp, Matthew Chantry, Ewan Pinnington, Margarita Choulga, Souhail Boussetta, Maria Kalweit, Joschka Bödecker, Carsten F. Dormann, Florian Pappenberger, and Gianpaolo Balsamo (2025). Advances in land surface forecasting: a comparison of LSTM, gradient boosting, and feed-forward neural networks as prognostic state emulators in a case study with ecLand. Geoscientific Model Development, 18(4), 921–937. doi:10.5194/gmd-18-921-2025 web

2024

  • Marieke Wesselkamp, Niklas Moser, Maria Kalweit, Joschka Boedecker, Carsten F. Dormann. Process-informed neural networks: a hybrid modelling approach to improve predictive performance and inference of neural networks in ecology and beyond. Ecology Letters. web
  • Paul Schmidt-Barbo, Gabriel Kalweit, Mehdi Naouar, Lisa Paschold, Edith Willscher, Christoph Schultheiss, Bruno Markl, Stefan Dirnhofer, Alexandar Tzankov, Mascha Binder und Maria Kalweit. Detection of disease-specific signatures in B cell repertoires of lymphomas using machine learning. In PLOS Computational Biology. web
  • Maria Kalweit and Gabriel Kalweit. Warum wir neu lernen müssen, mit Maschinen zu sprechen – eine Momentaufnahme der Generativen KI im Januar 2024. In Ordnung der Wissenschaft. web
  • Gabriel Kalweit, Anusha Klett, Mehdi Naouar, Jens Rahnfeld, Yannick Vogt, Diana Laura Infante Ramirez, Rebecca Berger, Jesus Duque Afonso, Tanja Nicole Hartmann, Marie Follo, Michael Luebbert, Roland Mertelsmann, Evelyn Ullrich, Joschka Boedecker and Maria Kalweit. Unsupervised Feature Extraction from a Foundation Model Zoo for Cell Similarity Search in Oncological Microscopy Across Devices. ICML 2024 Workshop on Foundation Models in the Wild. web
  • Hao Zhu, Brice De La Crompe, Gabriel Kalweit, Artur Schneider, Maria Kalweit, Ilka Diester and Joschka Boedecker. Multi-intention Inverse Q-learning for Interpretable Behavior Representation. TMLR, 2024. web

2023

  • Maria Kalweit, Andrea Burden, Joschka Boedecker, Thomas Hügle and Theresa Burkard. Patient groups in Rheumatoid arthritis identified by deep learning respond differently to biologic or targeted synthetic DMARDs. PLoS Computational Biology, 2023. web
  • Gabriel Kalweit, Maria Kalweit, Ignacio Mastroleo, Joschka Bödecker und Roland Mertelsmann. Künstliche Intelligenz in der Krebstherapie. Ordnung der Wissenschaft, 2022. web
  • Mehdi Naouar, Gabriel Kalweit, Ignacio Mastroleo, Philipp Poxleitner, Marc Metzger, Joschka Boedecker and Maria Kalweit. Robust Tumor Detection from Coarse Annotations via Multi-Magnification Ensembles. Poster at Digital Oncology, Hannover 2023. web
  • Mehdi Naouar, Gabriel Kalweit, Anusha Klett, Yannick Vogt, Paula Silvestrini, Diana Infante, Roland Mertelsmann, Joschka Boedecker and Maria Kalweit. CellMixer: Annotation-free Semantic Cell Segmentation of Heterogeneous Cell Populations. NeurIPS 2023 Workshop on Medical Imaging. web
  • Yannick Vogt, Mehdi Naouar, Maria Kalweit, Cornelius Miething, Justus Duyster, Roland Mertelsmann, Gabriel Kalweit and Joschka Boedecker. Stable Online and Offline Reinforcement Learning for Antibody CDRH3 Design. NeurIPS 2023 Workshop on Machine Learning in Structural Biology. web

2022

  • Maria Kalweit, Gabriel Kalweit, Moritz Werling and Joschka Boedecker. Deep Surrogate Q-Learning for Autonomous Driving. ICRA 2022.
  • Gabriel Kalweit, Maria Kalweit, Joschka Boedecker. Robust and Data-efficient Q-learning by Composite Value-estimation. TMLR 2022. web
  • Thomas Hügle, Leo Caratsch, Matteo Caorsi, Jules Maglione, Alexandre Dumusc, Diana Dan, Marc Blanchard, Gabriel Kalweit and Maria Kalweit. Automated Recognition and Monitoring of Dorsal Finger Folds by a Convolutional Neural Network as a Potential Digital Biomarker for Joint Swelling in Patients with Rheumatoid Arthritis. Digital Biomarkers, 2022. web

2021

  • Branka Mirchevska, Maria Hügle, Gabriel Kalweit, Moritz Werling, Joschka Boedecker. Amortized Q-learning with Model-based Action Proposals for Autonomous Driving on Highways. ICRA 2021. web
  • Maria Hügle, Ulrich A Walker, Axel Finckh, Ruediger Mueller, Gabriel Kalweit, Almut Scherer, Joschka Boedecker, Thomas Hügle. Personalized Prediction of Disease Activity in Patients with Rheumatoid Arthritis Using an Adaptive Deep Neural Network. PLOS ONE. web
  • Maria Kalweit, Gabriel Kalweit and Joschka Boedecker. AnyNets: Adaptive Deep Neural Networks for Medical Data with Missing Values. IJCAI 2021 Workshop on Artificial Intelligence for Function, Disability, and Health. web
  • Maria Kalweit, Gabriel Kalweit, Moritz Werling and Joschka Boedecker. Deep Surrogate Q-Learning for Autonomous Driving. IJCAI 2021 Workshop on Artificial Intelligence for Autonomous Driving.
  • Gabriel Kalweit, Maria Kalweit, Mansour Alyahyay, Zoe Jaeckel, Florian Steenbergen, Stefanie Hardung, Ilka Diester and Joschka Boedecker. NeuRL: Closed-form Inverse Reinforcement Learning for Neural Decoding. ICML 2021 Workshop on Computational Biology. web
  • Gabriel Kalweit, Maria Huegle, Moritz Werling and Joschka Boedecker. Q-learning with Long-term Action-space Shaping to Model Complex Behavior for Autonomous Lane Changes. IROS 2021. web
  • Hügle T, Kalweit M. Artificial intelligence-supported treatment in rheumatology: Principles, current situation and perspectives. Zeitschrift fur Rheumatologie. 2021 web

2020

  • Gabriel Kalweit*, Maria Huegle*, Moritz Werling and Joschka Boedecker. Deep Inverse Q-learning with Constraints. NeurIPS 2020. web
  • Maria Huegle, Gabriel Kalweit, Moritz Werling and Joschka Boedecker. Dynamic Interaction-Aware Scene Understanding for Reinforcement Learning in Autonomous Driving. In Proc. of the IEEE International Conference on Robotics and Automation (ICRA), 2020. web
  • Maria Hügle, Gabriel Kalweit, Thomas Hügle and Joschka Boedecker. A Dynamic Deep Neural Network For Multimodal Clinical Data Analysis. AAAI 2020 Workshop on Health Intelligence. Explainable AI in Healthcare and Medicine. Studies in Computational Intelligence, Springer (2020). web
  • Maria Hügle, Patrick Omoumi, Jaap van Laar, Joschka Boedecker and Thomas Hügle. Applied Machine Learning and Artificial Intelligence in Rheumatology. Rheumatology Advances in Practice (2020). web

2019

  • Maria Huegle*, Gabriel Kalweit*, Branka Mirchevska, Moritz Werling and Joschka Boedecker. Dynamic Input for Deep Reinforcement Learning in Autonomous Driving. IROS 2019. web
  • Maria Hügle, Joschka Boedecker and Thomas Hügle. Clinical Event-based Adaptive Deep Learning for the Prediction of Disease Progression in Rheumatoid Arthritis (Abstract and poster at Frontier of AI-Assisted Care Scientific Symposium, Stanford University, 2019).
  • Maria Hügle, Gabriel Kalweit, Moritz Werling, and Joschka Boedecker. Learning Dynamic Representations for Deep Reinforcement Learning in Autonomous Driving (Extended abstract at RSS 2019 Workshop on Scene and Situation Understanding for Autonomous Driving).
  • Gabriel Kalweit, Maria Huegle and Joschka Boedecker. Composite Q-Learning. Extended abstract at RSS 2019 Workshop on Combining Learning and Reasoning.

2018

  • Maria Hügle, Simon Heller, Manuel Watter, Manuel Blum, Farrokh Manzouri, Matthias Dümpelmann, Andreas Schulze-Bonhage, Peter Woias, Joschka Boedecker. Early Seizure Detection with an Energy-Efficient Convolutional Neural Network on an Implantable Microcontroller (IJCNN 2018, won the “Best Paper Award”). web
  • Simon Heller, Maria Hügle, Iman Nematollahi, Farrokh Manzouri, Matthias Dümpelmann, Andreas Schulze-Bonhage, Joschka Boedecker, Peter Woias. Hardware Implementation of a Performance and Energy-Optimized Convolutional Neural Network for Seizure Detection (EMBC 2018). web

2016

Molteni P, Hügle T, Hügle M, Nüesch C, Mündermann A. Reduction in ulnar pressure distribution when walking with forearm crutches with a novel cuff design: Cross-sectional intervention study on the biomechanical efficacy of an ulnar recess (2016).

Patents

  • Maria Huegle, Gabriel Kalweit, Branka Mirchevska, Moritz Werling and Joschka Boedecker. (EP3730369A1) Selecting a Motion Action for an Automated Vehicle Considering a Variable Number of Other Road Users.
  • Moritz Werling, Branka Mirchevska, Joschka Bödecker, Maria Hügle and Gabriel Kalweit. (10 2020 106 816.6) Steuerung eines automatisierten Kraftfahrzeugs.
  • Moritz Werling, Gabriel Kalweit, Maria Hügle and Joschka Bödecker. (10 2020 121 150.3) Trainieren eines Reinforcement Learning Agenten zur Steuerung eines autonomen Systems.
  • Joschka Bödecker, Maria Hügle, Gabriel Kalweit and Moritz Werling. (WO002022033746A1) Training a Reinforcement Learning Agent to Control an Autonomous System.
  • Joschka Boedecker, Maria Huegle, Gabriel Kalweit, Moritz Werling. (US 2024/0037447 A1) Training a Reinforcement Learning Agent to Control an Autonomous System.

Honors and Awards

  • Gips-Schüle-Nachwuchspreis in der Kategorie Technikwissenschaften, 2024
  • Wolfgang-Gentner-Nachwuchsförderpreis, Universität Freiburg, 2023
  • Best Paper Award, IJCNN 2018