There are many barriers to cross-cohort analyses and physically pooling data is constrained by governance, ethical and legal challenges. InterConnect is therefore taking a new approach. Rather than physically pooling data – i.e. ‘taking the data to the analysis’ – we are ‘taking the analysis to the data’.
Analytical instructions are sent remotely and the analysis is performed locally so all data stays at source and only results are shared. In this way, it is possible to perform an analysis that is equivalent to a meta-analysis of harmonised individual level data and so the approach is called ‘federated meta-analysis’. Because the data stays at source, it remains within the governance structure and control of the originating study.
- Read more about federated meta-analysis below.
- Read more about the privacy and security of the approach here
Federated meta-analysis provides a secure, scalable and sustainable approach to cross-cohort analyses through what could be called a ‘data access and results sharing’ network.
Federated meta-analysis
Data from contributing studies are held securely on local, study-based computers.
- Analytical commands are sent as blocks of code to request each local computer to undertake an analysis and return summary results that are non-identifying.
- Analyses are performed locally so all data stay at source and does not leave the originating institution.
- A number of measures are in place to ensure privacy and security.
- Read more detailed information on how analyses are performed and how data security are protected (pdf).