With colleagues from the ICAI Stroke Lab we are investigating potential use-cases for synthetic tabular data for prediction modelling. Specifically, we aim to mitigate class imbalance by generating synthetic samples for minority classes, thereby improving predictive performance.