"Federated Data Networks is a concept that allows stand alone data sources to create a virtual view of their combined information. Data sources store their data in the same format with standardized definitions so that queries run across all sites. A virtual database created by data federation software doesn't contain the data itself. Instead, it has information (metadata)about the data and its location. The actual data is left at its source.
To address security and custodian issues, FDN follows the principle of edge analytics. The assumption is that data at the atomic level is never passed to a centralized analytics process where a discovery tool scans atomic level data to create information for analysis.
The FDN allows a centralized or decentralized analytics process but does not store the atomic level data in a centralized resource. Rather, the analytics process has access to this data, managed and made available at the source or edge, but does not keep the source data or extract to generate a desired model.
To do this requires a basic understanding of structured query language (SQL). Queries are a form of calculus exploring the relationships among data attributes in a source.
Making a single call to multiple data sources, then integrating and organizing the data in a middleware layer is referred to as data virtualization: details on the data itself, and data management of source location, protocols and formats.
A Health Information Exchange (HIE) & Clinical Data Repository (CDR) managed by a HIE authority, are maintained by hospitals. Each hospital transmits patient health data to the CDR, where it is continually updated via a hospital health information system (HIS).
Each node in the federated network requires an authentication layer for every owner and data requestor. Typically nodes are organized as data "communities" that define who create or access data. Communities tend to organize around departments in a hospital. This allows for different permission rules to access and exchange information.
