Can Abstract Screening Workload Be Reduced Using Text Mining? User Experiences of the Tool Rayyan

Background: One time-consuming aspect of conducting systematic reviews is the task of sifting through abstracts to identify relevant studies. One promising approach for reducing this burden uses text mining technology to identify those abstracts that are potentially most relevant for a project, allowing those abstracts to be screened first.

Objectives: To examine the effectiveness of the text mining functionality of the abstract screening tool Rayyan. User experiences were collected.

Methods: Rayyan was used to screen abstracts for 6 reviews in 2015. After screening 25%, 50%, and 75% of the abstracts, the screeners logged the relevant references identified. A survey was sent to users.

Results: After screening half of the search result with Rayyan, 86% to 99% of the references deemed relevant to the study were identified. Of those studies included in the final reports, 96% to 100% were already identified in the first half of the screening process. Users rated Rayyan 4.5 out of 5.

Discussion: The text mining function in Rayyan successfully helped reviewers identify relevant studies early in the screening process.