Yesterday was the last day of a really interesting two-day symposium on automation and systematic reviews in Bristol. The main participants were computer scientists and systematic reviewers; I belonged in the relatively small ‘other’ group.
It struck me that the focus was on breaking down the steps of systematic reviews (as seen in a few papers, one reviewed on this blog – click here) and using automation techniques at each stage. It was during the course of the symposium that it became apparent that there are two main approaches to undertaking systematic reviews more quickly (AKA rapid reviews):
- Following the systematic review process and replacing humans with automation where possible.
- Mimicking the systematic review process but taking short cuts where possible e.g. only searching one database, not under-taking a meta-analysis.
(It doesn’t take too much imagination to see that you could use automation techniques on the second approach).
The third, more esoteric, approach I would suggest is the ‘other’ type, for instance Trip’s five minute systematic review method. This is focused on delivering an appropriate outcome as quickly as possible using a variety of heuristics (albeit in the Trip system these are powered by automation techniques).
To get a flavour of the symposium you can see the tweets via the very nice hashtag #autosynthesis.