Focus Area Synthetic data generation
Vision
Our vision is to fundamentally change the way medical data is used by driving the generation of high-quality synthetic data that faithfully replicates real-world clinical trial data. By using advanced AI architectures such as GANs, VAEs and Transformers, we aim for a future where data sharing is seamless, secure and privacy-compliant. Our goal is to facilitate access to valuable medical data while maintaining strict anonymization standards to drive innovation in healthcare AI. Our main mission is to develop and deploy sophisticated AI-based techniques to generate synthetic tabular data for medical datasets. By using cutting-edge models such as GANs, VAEs and Transformers, we synthesize data that not only protects patient privacy but also ensures the integrity and utility of the information. A critical aspect of our work is rebalancing data sets through intelligent oversampling to address data imbalance issues and improve the performance of AI algorithms. In doing so, we enable more accurate, reliable and equitable AI-based medical insights and promote advances in research, diagnostics and patient care.
Focus Areas
- Development and use of AI-based techniques to generate synthetic tabular data for medical data sets
- Use of state-of-the-art models such as GANs, VAEs and transformers for data synthesis
- Ensuring patient privacy and the integrity and usefulness of information
- Rebalancing data sets through oversampling to address issues of unbalanced data
Contact Person
© MF/Stephan Wiegand
Research fellow
NameMr Dr.-Ing. Markus Wolfien
Send encrypted email via the SecureMail portal (for TUD external users only).
Selected Projects
- SCaDS.AI — Creation of tabular data in medicine