Comment by enjoykaz

100% precision on the Claude Code run seems undersold. If you're building a systematic review, a false match (claiming paper X studied trial Y when it didn't) could corrupt your conclusions in ways a miss wouldn't. Would be curious if the authors have a domain reason for weighting recall over precision here.