Hands-on approach towards quantifiable and scientific useful data anonymization for large medical datasets
Dr. Patrick Stähli, Data Scientist, Insel Data Science Center. For more information, please visit the website.
Györgyi Hamvas, Department of Angiology, Inselspital, University Hospital Bern.
Prof. Iris Baumgartner, Department of Angiology, Inselspital, University Hospital Bern.
Prof. Dr. med. Alexander Leichtle, Directorate of Teaching and Research and Department of Clinical Chemistry, Inselspital, University Hospital Bern.
Biomedical research needs to be performed inevitably with health-related data originating from patients. Anonymization protects these patients against identification. In this project, we investigate a hands-on approach to anonymize a dataset from the University Clinic of Angiology by applying a robust data anonymization methodology where re-identification risk doesn’t depend on the background knowledge of an adversary trying to re-identify patients. Data anonymization, however, results in a loss of information that potentially lowers the statistical power of the dataset. We address this question by comparing the statistical outcome of the original and the anonymized dataset to balance power and anonymity.