Using Rayyan's Predictions Classifier (Rayyan Beta)

Rayyan's AI-powered Predictions Classifier (previously called: 5-star Relevance Ranking) helps reviewers accelerate their systematic reviews by learning patterns in decision criteria as reviewers make decisions and using that learning to rate the probability that an article will be included.

To use the Rayyan Predictions Classifier, follow the steps below:

1. Make at least 50 inclusion and/or exclusion decisions in your systematic review, with at least 5 decisions of both types (included and excluded). 

2. Click Compute Ratings

You will see the message: Ratings are being calculated for review [Review Name] while Rayyan's Artificial Intelligence engine calculates the likelihood of each article to be included in your literature review (5 stars is most likely).

Once it is finished, you will see the message: New ratings available. You'll also see thumbs up/down ratings, in the Ratings column at the right, ranging from 2 thumbs down - near zero likelihood of inclusion to 2 thumbs up- very high likelihood of inclusion, and ? for no recommendation.

3. You can now sort your review by least or most likely to be included by clicking on the sorting arrow besides the Filter by Inclusion facet. When sorting on the rating system, the order is precise - in other words, the exact calculation dictates the order while the visible stars show an approximation.

4. As you make more decisions, you can compute ratings again at any time. The more decisions you make, the more accurate the ratings will be - as Rayyan learns more from each decision.