DataSHIELD – New Directions and Dimensions

Abstract

<div class="page" title="Page 1"><div class="layoutArea"><div class="column"> <span>In disciplines such as biomedicine and social sciences, sharing and combining sensitive individual-level data is often prohibited by ethical-legal or governance constraints and other barriers such as the control of intellectual property or the huge sample sizes. DataSHIELD (</span><strong>D</strong><span>ata </span><strong>A</strong><span>ggregation </span><strong>T</strong><span>hrough </span><strong>A</strong><span>nonymous </span><strong>S</strong><span>ummary-statistics from </span><strong>H</strong><span>armonised </span><strong>I</strong><span>ndividual-lev</span><span><strong>EL</strong> <strong>D</strong></span><span>atabases) is a distributed approach that allows the analysis of sensitive individual-level data from one study, and the co-analysis of such data from several studies simultaneously without physically pooling them or disclosing any data. </span> <span>Following initial proof of principle, a stable DataSHIELD platform has now been implemented in a number of epidemiological consortia. This paper reports three new applications of DataSHIELD including application to post-publication sensitive data analysis, text data analysis and privacy protected data visualisation. Expansion of DataSHIELD analytic functionality and application to additional data types demonstrate the broad applications of the software beyond biomedical sciences. </span> </div></div></div>