The Evidence Synthesis Group helps researchers across the BRC apply state-of-the-art methods for systematic reviews, meta-analyses and evidence syntheses. We ensure that researchers across all themes use efficient and robust methods, such as minimising any biases and errors, and identifying opportunities for appropriate use of complex synthesis methods.
We develop methods for systematic reviews of study types for which current methods are less well developed than for Phase III trials. This includes development of frameworks to assess the validity of observational and other non-randomized studies, such as our Risk Of Bias In Non-randomised Studies (ROBINS) family of assessment tools.
Automation and efficiencies in systematic reviews (led by Professor Julian Higgins)
Developing novel interventions needs to be informed by existing research, but the amounts of relevant literature can be vast. To support multiple programmes of evidence reviews and syntheses, we adopt, adapt and further develop methods that seek to make the process more efficient, including text mining and machine learning methods. We work on methods to improve the processes of searching for relevant articles, mapping the existing evidence, and pulling out key pieces of information from articles.
Network meta-analysis and multi-parameter evidence synthesis (led by Professor Nicky Welton)
We have particular expertise in multi-parameter evidence synthesis methods, which answer research questions by piecing together findings from sources providing evidence on different, but mathematically connected, parameters. In this workstream, we apply and further develop these approaches, in order to build coherent models consistent with and informed by all the available evidence.