Data Access and Reproducibility in Privacy Sensitive eScience Domains

In privacy sensitive eScience domains, the data forming the basis for investigations is attributable for example to individuals. However, the disclosure of such data is often not allowed or advised if it contains sensitive data about the individual. Thus special attention needs to be paid when conducting eScience experiments, so that such data is not accessed in unauthorised ways. This affects the data in original or transformed forms, if the latter still allows deduction of information on individuals. Such concerns are opposing interests of repeatability and reproducibility, where the input data and traces of experiment executions form an important aspect to enable such goals. In this paper, we present a use case in the area of health policy planning, where statistical and mathematical models are trained from routine data health data, which contains privacy sensitive information.We thus discuss requirements for protecting the privacy, with the goal of still enabling repeatability and reproducibility.