aiqbench analysis
This article introduces TC3-VQA, a specialized dataset designed to train vision-language models for Tactical Combat Casualty Care by linking visual injury observations with established clinical doctrine. The resource enables researchers to evaluate how AI systems can recognize interventions, recall procedural guidance, and appropriately abstain when visual evidence is insufficient. By grounding answers in verbatim source passages, the dataset addresses critical needs for safety-critical medical applications where hallucinations are unacceptable.
A doctrine-grounded visual question answering dataset for Tactical Combat Casualty Care
arXiv:2610.07339v1 Announce Type: cross Abstract: Tactical Combat Casualty Care (TC3) requires responders to connect visual observations of injuries and interventions with established clinical guidance. Developing vision-language models to support this process requires supervision that links visible evidence to traceable doctrine. We present TC3-VQA, a dataset constructed from public instructional and field TC3 videos and authoritative TC3 documents. It contains 581 items spanning 11 concepts, with 1,860 questions covering intervention recognition, doctrine, clinical reasoning, procedural guidance, and refusal when visual information is insufficient. Doctrine-based answers preserve verbatim source passages and character offsets. Construction combines visual annotation, passage retrieval, entailment checks, and verification across model families. Equipment boxes, anatomical labels, temporal segments, and source metadata accompany the question-answer pairs. Automated audits and ratings by two physicians and two medical students characterize annotation quality, with human ratings available for 88 retained items. The dataset provides a resource for adapting vision-language models to TC3, studying the connection between visual evidence and clinical knowledge, and evaluating recognition, doctrine recall, and abstention.
Abstract:Tactical Combat Casualty Care (TC3) requires responders to connect visual observations of injuries and interventions with established clinical guidance. Developing vision-language models to support this process requires supervision that links visible evidence to traceable doctrine. We present TC3-VQA, a dataset constructed from public instructional and field TC3 videos and authoritative TC3 documents. It contains 581 items spanning 11 concepts, with 1,860 questions covering intervention recognition, doctrine, clinical reasoning, procedural guidance, and refusal when visual information is insufficient. Doctrine-based answers preserve verbatim source passages and character offsets. Construction combines visual annotation, passage retrieval, entailment checks, and verification across model families. Equipment boxes, anatomical labels, temporal segments, and source metadata accompany the question-answer pairs. Automated audits and ratings by two physicians and two medical students characterize annotation quality, with human ratings available for 88 retained items. The dataset provides a resource for adapting vision-language models to TC3, studying the connection between visual evidence and clinical knowledge, and evaluating recognition, doctrine recall, and abstention.
20 pages, 6 figures, 5 tables. Dataset: this https URL code: this https URL
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
arXiv:2610.07339 [cs.CV]
(or arXiv:2610.07339v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2610.07339 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 279 words
Related coverage
Game Design Creativity evals
LessWrong · Oct 12, 2026
Problems of Proliferation
LessWrong · Sep 29, 2026
DRAGON: A Benchmark for Evidence-Grounded Visual Reasoning over Diagrams
arXiv · Sep 30, 2026
A Red Line and Oversight Framework for Government AI Contracts
AI Alignment Forum · Jul 18, 2026