When unique records are entered into the database, they are ready to be screening on eligibility and to be classified.
We currently annotate, with the help of a ‘crowd’, the publications based on:
People who are confidently able to differentiate between different study designs are welcomed to join the ‘crowd’. When potential contributors are unsure they are able to do the work, they can take a test: here.
To be able to distribute screening tasks to a ‘crowd’, we build a shiny app that communicates with the central database. Records are attibuted to members of the crowd for screening. When the task is completed, the decisions are verified by a second member of the crowd. Disagreement is resolved by the coordinator or by a third crowd member (Figure 1).
Figure 1. The workflow of screening and verification tasks. Crowd members are represented by numbered hexagons.
The screening app offers a login-protected environment that communicates with the central database. Crowd members are presented with citation information and are asked to make several decisions. Their decisions are submitted to the database.
To accelerate the screening and verification of publications, we built a simple shiny app. This loads references and displays these to the reviewers. They execute their tasks in the shiny environment and decisions are logged into the central database. A demo of a simple screening app can be found here and the source code is available here.