Sum-of-Checks: Structured Reasoning for Surgical Safety with Large Vision-Language Models
Weiqiu You, Cassandra Goldberg, Amin Madani, Daniel A. Hashimoto, Eric Wong · 2026
Read Sum-of-Checks: Structured Reasoning for Surgical Safety with Large Vision-Language Models by Weiqiu You, Cassandra Goldberg, Amin Madani, Daniel A. Hashimoto, Eric Wong (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-22156.txt.
Description
You and four colleagues introduce Sum-of-Checks, splitting each safety criterion into expert-set reasoning checks. The checks reflect clinical visual evidence. Given a frame, a vision-language model judges each check with a binary verdict and justification, and scores use fixed, weighted aggregation. On Endoscapes2023, Sum-of-Checks beats direct prompting, chain-of-thought, and sub-question splitting, with and without few-shot examples. It lifts frame-level mean average precision by 12 to 14 percent over the best baseline.
Catalog record
| CREATOR | Weiqiu You, Cassandra Goldberg, Amin Madani, Daniel A. Hashimoto, Eric Wong |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 15536 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.22156