[论文] Beyond the Leaderboard: Design Lessons for Trustworthy Multimodal VQA

论文概要

研究领域: cs.CL, cs.CV 作者: Sushant Gautam, Vajira Thambawita, Michael A. Riegler, Pål Halvorsen, Steven A. Hicks 发布时间: 2026-07-16 arXiv: 2607.15241

中文摘要

医疗多模态 AI 必须同时融合视觉和文本证据，同时保持可靠性和可解释性。本文以 MediaEval Medico 2025 为回顾性 GI 内镜案例研究，分析了九个已记录系统在问答和解释质量方面的设计选择。预训练骨干网络的参数高效适配在挑战中提供了强劲性能，但答案层面的提升并不总能转化为忠实且完整的临床推理。强制执行结构化推理和显式证据 grounding 的方法在异构问题类型中表现出更可靠的行为，尽管证据是相关性的而非基于消融实验的。这些结果推动了超越词汇重叠的评估、标准化的证据链接解释、防泄漏的数据治理，以及轻量级的鲁棒性和校准检查。研究结论支持基于数据融合、可解释性和弹性评估的可信赖医疗多模态 AI。

原文摘要

Healthcare multimodal AI must combine visual and textual evidence while remaining reliable and interpretable. Using MediaEval Medico 2025 as a retrospective GI endoscopy case study, we analyze design choices across nine documented systems for question answering and explanation quality. Parameter-efficient adaptation of pretrained backbones provides strong challenge performance, but answer-level gains do not consistently translate into faithful and complete clinical reasoning. Methods enforcing structured reasoning and explicit grounding show more reliable behavior across heterogeneous question types, although the evidence is correlational rather than ablation-based. These results motivate evaluation beyond lexical overlap, standardized evidence-linked explanations, leakage-aware data governance, and lightweight robustness and calibration checks. The findings support trustworthy multimodal healthcare AI based on data fusion, explainability, and resilient evaluation.

--- *自动采集于 2026-07-20*

#论文 #arXiv #AI #小凯