MAGIC：用基尼系数检测医疗多模态模型的幻觉
MAGIC: Modality-Aware Gini Inspection for Clinical Hallucination Detection in LMMs
医疗多模态模型看片子也会一本正经地胡说，这篇给了一个不用重训模型的检测法：用基尼系数看注意力在图像和文字间怎么分布，幻觉回答的分布明显不同。做医疗 AI 可靠性或 LLM 可解释性的研究者和工程团队都值得跑一下代码对比现有基线。
一篇 arXiv 论文提出 MAGIC（Modality-Aware Gini Inspection），把医疗 LMM 的幻觉检测建模为生成答案 token 时模态间注意力权重分布失衡的问题。方法上，作者用基尼系数量化各层、各注意力头中视觉 token 与文本 token 的注意力集中度，发现幻觉回答中视觉与文本 token 的注意力质量存在统计学显著的重分配。
基于这些模态注意力特征，作者训练了一个 logistic regression 分类器，在 2 个医疗 LMM、3 种医学影像模态上评测，成绩达到或超过现有幻觉检测基线。论文指出，现有检测方法多面向单模态场景，无法捕捉医学影像中视觉-语言的交互动态。
MAGIC: Modality-Aware Gini Inspection for Clinical Hallucination Detection in LMMs
Recent advances in medical large multimodal models (LMMs) have brought promising progress to clinical decision support, yet hallucination, the generation of confident but clinically unverifiable outputs, poses a critical threat to the reliability demanded in high-stakes medical settings. Existing hallucination detection methods, predominantly designed for unimodal settings, fail to capture the visual-linguistic dynamics inherent to medical imaging. We formulate the problem of hallucination detection in LMMs as disproportionate distribution of attention weights across the modalities during answer token generation. We quantify modality-specific attention concentration across layers and heads using the Gini coefficient and demonstrate statistically significant redistribution of attention mass between visual and textual tokens in hallucinated responses. We leverage these features to train a logistic regression-based classifier and evaluate it on two medical LMMs across three medical imaging modalities, demonstrating that our method achieves competitive or superior performance against contemporary hallucination detection baselines.
本条情报归属 AITOP「论文」核心追踪栏目。涉及核心实体与技术议题包括：LMM、幻觉检测、医学影像、基尼系数，读者可通过下方标签进入对应专题页查阅历史脉络与关联对比。
信源核验说明：本文原始线索采编自「arXiv cs.AI」（@Sachin Acharya, Niyoj Oli, Shishir Adhikari, Aminu Lawal, Yash Raj Shrestha, Prashnna K Gyawali, Binod Bhattarai），经 AITOP 多维评分管线结构化提炼。