aiqbench analysis
ClinHallu introduces a critical benchmark for diagnosing hallucinations in medical multimodal large language models by dissecting errors into specific reasoning stages like visual recognition and knowledge recall. This granular approach allows researchers to pinpoint exactly where failures occur within the model's decision pipeline, moving beyond simple accuracy metrics to address reliability in high-stakes clinical applications. By providing a testbed for trace-supervised fine-tuning, it offers actionable insights for improving the robustness of AI systems designed for medical diagnostics.
ClinHallu: A Benchmark for Diagnosing Stage-Wise Hallucinations in Medical MLLM Reasoning
arXiv:2606.14697v1 Announce Type: cross Abstract: Building trustworthy medical multimodal large language models (MLLMs) is critical for reliable clinical decision support. Existing medical hallucination benchmarks mainly focus on data collection, but often ignore where hallucinations originate within the reasoning process. We find that hallucination sources vary across samples: errors may arise from visual misrecognition, incorrect medical knowledge recall, or flawed reasoning integration. To enable source-level hallucination diagnosis, we introduce ClinHallu, a benchmark for stage-wise hallucination diagnosis in medical MLLM reasoning. ClinHallu contains 7,031 validated instances, where each instance is augmented with a structured reasoning trace decomposed into Visual Recognition, Knowledge Recall, and Reasoning Integration. We also use stage-replacement interventions to measure how correcting specific stages affects the final answer. Beyond evaluation, we show that trace-supervised fine-tuning reduces stage-wise hallucinations. ClinHallu provides a fine-grained hallucination testbed for diagnosing and mitigating reasoning failures in medical MLLMs. The benchmark is publicly available at https://github.com/alibaba-damo-academy/ClinHallu.
Abstract:Building trustworthy medical multimodal large language models (MLLMs) is critical for reliable clinical decision support. Existing medical hallucination benchmarks mainly focus on data collection, but often ignore where hallucinations originate within the reasoning process. We find that hallucination sources vary across samples: errors may arise from visual misrecognition, incorrect medical knowledge recall, or flawed reasoning integration. To enable source-level hallucination diagnosis, we introduce ClinHallu, a benchmark for stage-wise hallucination diagnosis in medical MLLM reasoning. ClinHallu contains 7,031 validated instances, where each instance is augmented with a structured reasoning trace decomposed into Visual Recognition, Knowledge Recall, and Reasoning Integration. We also use stage-replacement interventions to measure how correcting specific stages affects the final answer. Beyond evaluation, we show that trace-supervised fine-tuning reduces stage-wise hallucinations. ClinHallu provides a fine-grained hallucination testbed for diagnosing and mitigating reasoning failures in medical MLLMs. The benchmark is publicly available at this https URL.
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
arXiv:2606.14697 [cs.CV]
(or arXiv:2606.14697v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2606.14697 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 · 251 words
Related coverage
Lingshu: A Generalist Foundation Model for Unified Multimodal Medical Understanding and Reasoning
arXiv · Sep 22, 2026
MedPAIR: Measuring Whether Physicians and AI Agree on What Matters in Medical QA
Snorkel AI Blog · Oct 2, 2026
Introducing NV-Reason-CT Open 3D CT VLM for Radiologist Chain-of-Thought Reasoning
NVIDIA · Sep 23, 2026
Hallucination in Medical Imaging AI: A Cross-Modality Analytical Framework for Taxonomy, Detection, and Mitigation under Regulatory Constraints
arXiv · Jun 12, 2026