Electronic patient records are treasure troves of data that can be used for various types of analysis aimed at improving health care. However, the data also contains lots of sensitive information that should be excluded in certain analytical tasks (especially when sharing the results and/or the data with third parties). Therefore some level of anonymisation is almost always required.
The goal of this thesis is to explore the state of the art in anonymisation of electronic health records, with special emphasis on unstructured data (e.g., clinical notes). A proof-of-concept system implementing a selection of anonymisation techniques will then be developed, based on requirements of clinical professionals. The tool will be deployed on a sample of real clinical data (either acquired from a local hospital, or from publicly available resources like MIMIC or TCGA). The results will be described in the form of a thesis.