Privacy and data protection in cancer genomics are increasingly pressing concerns, yet the risks associated with clinical tumor profiling remain poorly understood. While genetic data is known to enable patient re-identification, specific variants which are essential for diagnosis and treatment, are still widely assumed to be safe. However, imperfect filtering pipelines, incomplete reference databases, and tumor heterogeneity might leak sensitive personal information, challenging current practices in oncology and research.
AURA-CTP addresses this gap by combining expertise from pathology, bioinformatics, and privacy research. The project focuses on two primary research goals: (i) systematically analyzing whether tumor data can be exploited for patient re-identification, and (ii) evaluating how such privacy risks would affect clinical practices and data-sharing policies. To achieve this, the team will (1) prepare genomic datasets from both public sources and clinical diagnostics, (2) adapt and test attack methods, (3) quantify privacy risks, (4) assess clinical implications, and (5) develop best-practice recommendations for secure data handling.
Collaboration and Funding: AURA-CTP is an interdisciplinary effort at RWTH Aachen University, uniting the Machine Learning in Cancer Genetics group, the Institute of Pathology, and the Communication and Distributed Systems group.
Involved PIs
- Klaus Wehrle
COMSYS, RWTH Aachen University - Kjong Lehmann Machine Learning in Cancer Genetics, RWTH Aachen University
- Nadina Ortiz-Brüchle
Institute for Pathology, RWTH Aachen University Hospital