Screen, Interact, Learn , Improve
ASReview is a machine learning tool for abstract screening. Users classify a subset of studies as relevant or irrelevant. This subset is used to train an algorithm that identifies features distinguishing relevant from irrelevant studies. The algorithm predicts the relevance of all studies and presents the most relevant study first. Users review and classify each presented study, with their decisions feeding back into the algorithm for continuous learning. This iterative process continues until the user decides to stop or all studies are classified. ASReview does not provide explicit stopping guidelines, posing a challenge in determining when to end the review and therefore the user has to pre-defined the stopping rules ( criteria that determine when to stop screening articles based on certain conditions). ASReview has the potential to save an average of about 60.2% of screening time. Potential time savings between 49 and 59 hours could be achieved, depending on the stopping rule.
