The sensitivity of health data requires special protection, and a number of technical and organizational procedures have been put in place within the AP-HP's Health Data Warehouse (HDW) to protect patients' personal data as effectively as possible, while enabling research projects to be carried out. Among these procedures, pseudonymization (removal of directly identifying data) and minimization (limiting data to what is necessary for the research) of the data studied reinforce this protection, but they are not neutral as they can impact the quality of the analyses carried out by the research teams. The aim of this project is to evaluate and optimize pseudonymization/minimization methods for different epidemiological study situations, in order to adapt them to the context of research on massive health data.