Breaking01 / AI
New rule cuts false safety approvals in gallbladder surgery
The new rule cut false approvals in videos where experts disagreed, but also approved fewer videos on time when they met the safety standard.

Researchers reported on 19 Nov 2026 in the preprint When accuracy hides risk that a new rule reduced false safety approvals in a comparison of gallbladder surgery videos (n=300, evaluation procedure set in advance), with peer review pending and independent replication still unconfirmed.
Using SAGES CVS Challenge videos, the researchers assessed whether the critical view of safety (CVS), an anatomical safety check, had been achieved before the first surgical clip was applied. The new rule uses multiple statistical tests to set a threshold that limits false approvals.
The paper reports that the standard rule falsely approved 8 videos and the new rule falsely approved 2 (n=39, videos on which expert reviewers disagreed). For the new rule's false approval rate, the one-sided 95% upper confidence bound was 0.15 (n=39, same disputed videos).
The researchers reported that the share approved on time fell from 35% under the standard rule to 10% under the new rule (videos that met the safety standard, comparison of both rules).
According to the paper, the standard rule sets its threshold using the overall average false approval rate. The new rule also sets limits for videos on which experts disagreed and for groups of videos by vendor and country.
The abstract leaves the new rule's false approval rate during live surgery unreported.