Using Maximum Entropy to Extend a Consent Privacy Impact Quantification

Due to the progress of digitization in the medical sector digital consent becomes more and more common. While digital consent itself has a huge number of benefits for the researcher it can impose a lot of questions for the individual giving it. One of those questions is what impact the consent to sharing data with a research project has on the individual's privacy. The Consent Privacy Impact Quantification (CPIQ) provides a quantification to help the user making a consent decision based on the potential data sharing risk and his individual acceptance preferences for a research project. While this quantification provides a good first estimation it has some limitations especially in the method the re-identification risk is calculated for a member of a dataset. This paper presents a method using the Maximum Entropy principle. This principle provides a way to measure the maximum unbiased distribution using limited background knowledge, which is provided by epidemiological data. This distribution can then be used to see how much higher the re-identification risk based on a sensitive attribute is compared to the uniform distribution. In addition, the first promising results of the method will be shown based on an experimental setting.