DILiS - News and Outreach Activities

May 29, 2026: Una Europa General Assembly in Paris

Data Science and AI are central topics for European research and innovation. At the Una Europa general assembly in beautiful Paris, I really enjoyed learning and discussing how international researchers around the chair Jussi Kangasharju from the SSC Data Science and AI envision improving knowledge-sharing, education and fostering cultural exchange in this fast-developing field. It was also great meeting, discussing and joining forces with Claudia Müller-Birn and Christian von Sikorski from Freie Universität Berlin.

May 18, 2026: Simulation-Based Transfer Learning Paper Published in BMC Bioinformatics: Selecting synthetic data for successful simulation-based transfer learning in dynamical biological systems

Machine learning is powerful, but in biology, training data is often scarce. Our new piece of research addresses a key question in ML for biology: When you do not have enough real-world data, can you train on synthetic data instead? Short answer: Yes, but it is only helpful if that synthetic data based on mechanistic prior knowledge is carefully designed and suits the task. Simon Witzke, Julian Zabbarov, Maximilian Kleissl, Pascal Iversen, Bernhard Renard, and Katharina Baum developed a systematic pipeline for selecting and evaluating the right size, diversity, and noise level for ODE-based synthetic datasets for transfer learning. Find the full paper published at BMC Bioinformatics: https://lnkd.in/duBMgg6e

May 12, 2026: DrEval Paper Published in Nature Communications: Critical evaluation of drug response prediction models with DrEval

ML-based cancer cell line drug response models are well-motivated, and significant research effort has gone into developing complex modeling approaches (over 100 papers in 2025). The problem: under rigorous evaluation, none of them actually works: most are published based on inflated metrics, break down when probed in realistic application scenarios, and are outperformed by simple baselines. Judith Bernett and Pascal Iversen, together with collaborators Mario Picciani, Katharina Baum, Markus List, and Mathias Wilhelm, built DrEval, a pipeline for unbiased evaluation of these models, to encourage more meaningful progress in the field. Now published in Nature Communications: https://lnkd.in/dPuynXD3. DrEval could also function as an unbiased reward signal for AI agent–based model development in biological ML. The team is happy to chat about this if anyone is interested in details.

February 17, 2026: Successful PhD Defense of Flavio Morelli

Huge congratulations to Flavio Morelli for successfully defending his PhD thesis, on Multi-omics integration for advancing ML-driven drug discovery! We are very proud of his achievement and wish him all the best for his future work at Bayer.

June 12-13, 2025: What's graphs got to do with this?

Katharina Baum gave a keynote talk at the GMDS Workshop Computational Models in Biology and Medicine that is hosted this year by Tim Kacprowski's group at TU Braunschweig. In the talk with the title "What's graphs got to do with it? Exploiting networks for predictions in the life sciences", she reported on three DILiS projects where biological graphs play a central role for prediction. Definitely a topic to follow up on!