PrizmaticQuestions we get
Common questions.
Straight answers, including where the honest answer is "not yet." If your question isn't covered here, write to us and we'll add it.
Can I see why a specific paper was excluded?
Yes, per record, with the stage and the reason. The full record history also downloads as a single file so a reviewer, an editor or a regulator can reconstruct any decision without asking you for it.
Which risk-of-bias instrument do you use?
Both. The pipeline classifies study design first and forks, so trials are assessed with RoB 2 and observational studies with ROBINS-I. Domain-level judgments come with the supporting quotes rather than a single overall verdict.
What happens when the model gets something wrong?
You find it, because every decision is logged against the record it applies to and every extracted figure carries the sentence it came from. Screening is where errors matter most, so the borderline band can be routed to a human before it propagates. How often it happens is a number we don't have yet: the validation run that would produce an honest one is still open work, and we would rather publish nothing than an estimate.
Is the output actually PRISMA compliant?
The flow diagram is generated from the execution graph itself, so the counts in your paper are the counts the pipeline actually produced. Compliance is not automatic. PRISMA is a reporting standard for a review a human is accountable for, and Prizmatic produces the record that lets you meet it, not the accountability.
Can I use this for a registered or published review?
People do. You remain the author and the guarantor, which means reviewing what the pipeline produced rather than forwarding it. The audit trail exists precisely so that review is possible in hours rather than by redoing the work.