Explainable AI for Clinical Omics Data Interpretation
Current AI models in pathology are often based on supervised learning requiring large amounts of training data and therefore perform well only on common diseases.
However, for usage in the clinic, the model also has to detect rare findings and be transparent to the pathologist.
This project investigates methods to detect rare findings and improve the robustness of the AI model, while also being explainable.
We develop AI-models for omics and image data as well as their multimodal integration paving the way towards successful integration into clinical practice.
Research Group Lead / Charité
Director
Doctoral Researcher