TAVR-VLM: Risk-Conditioned Causal Grounding for Hallucination-Resistant Report Generation
On the map: LLM reasoning → Hallucination detection and mitigation
Abstract (arXiv)
Transcatheter Aortic Valve Replacement (TAVR) planning requires meticulous multimodal reasoning. However, adapting Multimodal Large Language Models (MLLMs) to this high-stakes domain is severely impeded by diagnostic hallucinations, where generated text lacks anatomical grounding. To address this, TAVR-VLM is introduced: a novel framework featuring Risk-Conditioned Causal Grounding Attention (R-CGA) that instantiates a model-internal ``Risk rightarrow Region rightarrow Word'' structural grounding pathway. R-CGA compresses multimodal inputs into a causal risk bottleneck, purifying dense visual features into a global risk mask. During autoregressive generation, a support-projected causal consistency objective constrains token-level grounding within the risk-defined support mask. Evaluated on M3TAVR, a comprehensive 1,482-patient cohort, TAVR-VLM establishes a new state-of-the-art. It achieves an AUROC of 0.896, boosts CIDEr to 0.936, and drastically reduces the hallucination rate to 8.1%, thereby improving interpretability for evidence-based surgical AI.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Hallucination detection and mitigation, one of 17 directions in LLM reasoning. Builds on 15 of its 22 references that are in the map. First posted on arXiv 25 Jun 2026.
Builds on
15 of 22 references.
References · 15
- 2016XGBoost: A Scalable Tree Boosting Systemcited by 5,166
- 2014CIDEr: Consensus-based Image Description Evaluationcited by 2,046
- 2023A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questionscited by 1,601
- 2025InternVL3.5: Advancing Open-Source Multimodal Models in Versatility, Reasoning, and Efficiencycited by 1,510
- 2023LLaVA-Med: Training a Large Language-and-Vision Assistant for Biomedicine in One Daycited by 1,305
- 2019TabNet: Attentive Interpretable Tabular Learningcited by 648
- 2020Generating Radiology Reports via Memory-driven Transformercited by 368
- 2023METransformer: Radiology Report Generation by Transformer with Multiple Learnable Expert Tokenscited by 101
- 2021Cross-Attention is All You Need: Adapting Pretrained Transformers for Machine Translationcited by 74
- 2024R2Gen-Mamba: A Selective State Space Model for Radiology Report Generationcited by 8
- 2025DeepGB-TB: A Risk-Balanced Cross-Attention Gradient-Boosted Convolutional Network for Rapid, Interpretable Tuberculosis Screeningcited by 6
- 2025Causal-SAM-LLM: Large Language Models as Causal Reasoners for Robust Medical Segmentationcited by 5
3 more
- 2026SkinCLIP-VL: Consistency-Aware Vision-Language Learning for Multimodal Skin Cancer Diagnosiscited by 3
- 2026Dialectic-Med: Mitigating Diagnostic Hallucinations via Counterfactual Adversarial Multi-Agent Debatecited by 2
- 2026Semantic-Topological Graph Reasoning for Language-Guided Pulmonary Screeningcited by 1
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2026VGS-Decoding: Visual Grounding Score Guided Decoding for Hallucination Mitigation in Medical VLMs
- 2025Process Reward Models for Sentence-Level Verification of LVLM Radiology Reports
- 2025UniVRSE: Unified Vision-conditioned Response Semantic Entropy for Hallucination Detection in Medical Vision-Language Models
- 2026Deterministic Hallucination Detection in Medical VQA via Confidence-Evidence Bayesian Gain
- 2026Synergistic Perception-Reasoning Governance: Grounding Medical MLLMs with Verifiable Anatomical Evidence
- 2026VIHD: Visual Intervention-based Hallucination Detection for Medical Visual Question Answering
- 2025Causally-Grounded Dual-Path Attention Intervention for Object Hallucination Mitigation in LVLMs
- 2025Anatomical Region-Guided Contrastive Decoding: A Plug-and-Play Strategy for Mitigating Hallucinations in Medical VLMs