DataSHIELD Implementation Workshop
June 24, 2011
The DataSHIELD implementation workshop has been successfully held at the IPRI headquarters in Lyon.
A joint collaboration of:
DataSHIELD (Data aggregation through anonymous summary-statistics from harmonized individual level databases) is a new concept that allows for individual level meta-analysis without sharing the actual data between the different centres involved in a certain study. DataSHIELD was developed by Paul Burton (university of Leicester, UK), Philippe Laflamme and Vincent Ferretti (Public Population Project in Genomics (P3G), Montreal, Canada), as an interface module between the OPAL database application for biobanks and the R software.
DataSHIELD is an approach that circumvents the conflict in ‘competing public goods’ as it enables the conduct of statistical analysis for research projects without sharing data but only non-identifying summary parameters between computers. This has the advantage of not disclosing individual level information and easy manipulation of very large databases (such as the genetic database) as it does not require actual physical transfer of data.
Please refer to the following article for more details on the method:
The workshop consisted of demonstrations of DataSHIELD.
A first session was run for setting up DataSHIELD and undertaking a pooled analysis using DataSHIELD on dummy data. The second session consisted of implementing dataSHIELD on real data provided by IPRI. These data correspond to a multicentre case-control study on melanoma and have already been analysed and published: A multicentre epidemiological study on sunbed use and cutaneous melanoma in Europe. Bataille V et al. The results from the dataSHIELD implementation have been successfully validated with the original published results.
DataSHIELD is presently being considered as the preferred approach for several worldwide projects.