Research Data Sharing Models: Examining Architecture and Governance

With the increased importance in data sharing that we are seeing in many areas of science, researchers and institutions need to be equipped with a better understanding of data sharing models that can help them to make informed decisions about their data sharing strategy. This paper will clarify the concept of data sharing models for sharing biomedical and behavioral research data. Each model highlighted in this analysis, including centralized, distributed, federated, and mixed models, will be described based on key features in its architecture and governance. To further highlight the similarities and differences between these models, these features are summarized in a single, quick-reference figure. In our own research with these models, we found the lack of clarity around their descriptions and differences to be frustrating. The goal of this analysis is to compile information about these models, that is typically high-level or does not appear in the same article, into a digestible and aggregated resource. Data sharing policies are making it imperative that researchers and institutions craft a plan for information sharing in the early stages of their research process. For some, this has challenging implications, including cost and training. Having a defined framework to build this around can help to conceptualize and ease this process. This resource will provide a great foundational understanding of the landscape of popular data sharing models to assist these efforts.