A must-read for everyone working with or interested in synthetic clinical trial data.
This work, part of Waldemar Hahn’s PhD project, shows that compound-metric hyperparameter optimization (HPO), together with robust data pre- and postprocessing, and dedicated privacy audits is key to generating high-quality, trustworthy synthetic clinical datasets.
Read the full paper here:
https://www.sciencedirect.com/science/article/pii/S0020025525010631
