Guest lecture by Dr Sagar Sen in Teams on Monday 14.15
Title: Anonymization and Re-identification Risk Analysis at the Cancer Registry of Norway
Abtract: The Cancer Registry of Norway curates data from its screening programs, cancer cases, deaths due to cancer, and corresponding health records to perform epidemiological research in Norway. The registry internally publishes scientific articles and annual reports about the epidemiology of cancer in Norway. The registry is also required to share de-identified data with external researchers who combine personal cancer data with other sources of data such as socio-economic factors of individuals from Statistical Bureau to generate novel correlations of interest. Privacy of individual identity is an essential ingredient in these activities as highly sensitive data needs to de-identified, shared and combined with other data sources. In this talk, I will describe my experience at the Cancer Registry of Norway where we developed an algorithm to increase privacy protection in personal data while maintaining a level of scientific utility. The presentation will talk about anonymization criteria (k-anonymization, l-diversity, t-closeness) that we needed to take into consideration, different types of attacks on privacy such the prosecutor attack, re-identification risk analysis. Students will get a demo of re-identification risk analysis using the tool ARX that is also used by many public sector companies in Norway.
References:
Ursin, Giske, Sagar Sen, Jean-Marie Mottu, and Mari Nygård. "Protecting privacy in large datasets—first we assess the risk; then we fuzzy the data." Cancer Epidemiology and Prevention Biomarkers 26, no. 8 (2017): 1219-1224.