A computer system utilizes a dataset to support a research study. One or more regions of interestingness are determined within a model of a first set of data records that are authorized for the research study by associated entities. A second set of data records is represented within the model, wherein the second set of data records are relevant for supporting objectives of the research study after de-identification. Records from the second dataset that are particularly useful for supporting objectives of the research study are identified, and authorization is requested from the corresponding entities of the identified data records from the second set of data records. After receiving authorization, those records are included with the first set to generate a resulting dataset. Embodiments of the present invention further include a method and program product for processing requests for health information in substantially the same manner described above.