KI and Multi-Omics Data

Background 

Nowadays, machine learning systems are about to expand into every area of our lives, including extremely sensitive domains such as medicine. Thus, finding ways to foster justified trust in machine learning models becomes increasingly essential, e.g., by identifying and implementing properties that render these methods trustworthy. 
To address this challenge, an interdisciplinary approach is necessary including perspectives from medicine, biology, computer science mathematics, and ethics. 

This group is affiliated with the Lower Saxony Centre for AI and Causal Methods in Medicine (CAIMed), which provides a collaborative framework for realizing such complex, cross-disciplinary research projects.
 

Group Leader: Prof. Dr. Kerstin Lenhof 

Research Foci - Goals

The Integrative Bioinformatics group develops trustworthy machine learning methods for medical applications, with a particular focus on oncology. In our research, we place emphasis on the reliability, robustness, and interpretability of machine learning models, alongside ethical considerations and human factors relevant to clinical deployment. We aim to ensure that these technologies align with societal needs and ethical standards.

A central question we address is how to generate trustworthy recommendations for anti-cancer drug treatments. While oncology is our primary application domain, the methods we develop are broadly applicable: we work on conformal prediction, multi-omics integration strategies, and transfer learning. We are also highly interested in conducting interview studies and surveys to understand the needs of stakeholders, e.g., medical doctors or patients. Currently, we are also exploring how drug recommendation methods from oncology can be adapted to other areas, such as predicting bacterial resistance.

Current Projects

• Trustworthy anti-cancer drug treatment prioritization

• Transfer learning from bulk cell lines to single cells 

• Bacterial resistance prediction

Selected Publications

Eckhart, L., Lenhof, K., Herrmann, L., Rolli, L. M., & Lenhof, H. P. (2025). How to predict effective drug combinations–moving beyond synergy scores. iScience, 28(6).

Lenhof, K., Eckhart, L., Rolli, L. M., Volkamer, A., & Lenhof, H. P. (2024). Reliable anti-cancer drug sensitivity prediction and prioritization. Scientific Reports, 14(1), 12303.

Lenhof, K., Eckhart, L., Rolli, L. M., & Lenhof, H. P. (2024). Trust me if you can: a survey on reliability and interpretability of machine learning approaches for drug sensitivity prediction in cancer. Briefings in Bioinformatics, 25(5), bbae379.

 

Members of the group