datashield-tidyverse

DataSHIELD is a software infrastructure and set of R packages designed to enable secure, privacy-preserving analysis of sensitive research data. It addresses the challenge of conducting collaborative data science without compromising individual privacy, particularly in the context of health data. The platform is built around privacy by design principles, ensuring that analysts do not directly access or view individual-level data, and that such data is never shared or moved from its original site.

Key features of DataSHIELD include support for federated analysis, allowing aggregated or global analyses to be performed across multiple data sites simultaneously, all without disclosing individual patient information. The analytic functions within DataSHIELD are designed to only share non-disclosive summary statistics, with automated output checking based on statistical disclosure control. Data sites themselves set the thresholds for these output checks, further reinforcing local control over data privacy. The base package, dsBase, undergoes extensive software validation, running over 12,000 tests per release to ensure correct operation and compliance with disclosure control requirements.

DataSHIELD is open source and has been developed by OBiBa and Molgenis, with ongoing maintenance and contributions from a community that includes members from academia, healthcare, industry, and small to medium-sized enterprises. The software is positioned for analysts, data custodians, and developers working with sensitive or consented research data who require robust privacy protections and compliance with best practice data governance.

The open source nature of DataSHIELD supports transparency and collaborative development. The platform has been in use for over a decade, demonstrating its established role in secure data science collaboration.