Fig. 4: Use cases of synthetic data.

Privacy conservation of synthetic data alleviates concerns of sharing identifying patient data. This enables simplified data sharing with the broader scientific community. Large publicly available data sets enable exploratory analysis of the synthetic data itself or by using synthetic data to augment proprietary data sets. This may encompass analyses of the genetic landscape of AML or the evaluation of the impact of specific alterations on patient risk. Further, training machine learning models requires large data sets. Training on publicly available synthetic data and validating on real data sets or vice versa, training on real data and using synthetic data as a benchmarking set for validation may foster the development of more robust machine learning models. Lastly, augmenting clinical trial cohorts with synthetic data or even substituting control cohorts with synthetic data entirely could re-shape prospective clinical trial designs. This, however, requires external validation and diligent regulatory oversight prior to implementation.