MedCausalX: Adaptive Causal Reasoning with Self-Reflection for Trustworthy Medical Vision-Language Models

TL;DR

MedCausalX introduces an adaptive causal reasoning framework for medical vision-language models, enhancing diagnostic accuracy and reliability by explicitly modeling causal chains and verifying reasoning through a novel two-stage reflection architecture.

Contribution

The paper presents MedCausalX, a novel end-to-end framework that explicitly models causal reasoning in medical VLMs, including a new dataset, a two-stage reflection architecture, and a trajectory-level causal correction method.

Findings

Outperforms state-of-the-art methods in medical diagnosis tasks.

Improves diagnostic consistency by +5.4 points.

Reduces hallucination in model outputs by over 10 points.

Abstract

Vision-Language Models (VLMs) have enabled interpretable medical diagnosis by integrating visual perception with linguistic reasoning. Yet, existing medical chain-of-thought (CoT) models lack explicit mechanisms to represent and enforce causal reasoning, leaving them vulnerable to spurious correlations and limiting their clinical reliability. We pinpoint three core challenges in medical CoT reasoning: how to adaptively trigger causal correction, construct high-quality causal-spurious contrastive samples, and maintain causal consistency across reasoning trajectories. To address these challenges, we propose MedCausalX, an end-to-end framework explicitly models causal reasoning chains in medical VLMs. We first introduce the CRMed dataset providing fine-grained anatomical annotations, structured causal reasoning chains, and counterfactual variants that guide the learning of causal relationships beyond superficial correlations. Building upon CRMed, MedCausalX employs a two-stage adaptive reflection architecture equipped with $\langle$causal$\rangle$ and $\langle$verify$\rangle$ tokens, enabling the model to autonomously determine when and how to perform causal analysis and verification. Finally, a trajectory-level causal correction objective optimized through error-attributed reinforcement learning refines the reasoning chain, allowing the model to distinguish genuine causal dependencies from shortcut associations. Extensive experiments on multiple benchmarks show that MedCausalX consistently outperforms state-of-the-art methods, improving diagnostic consistency by +5.4 points, reducing hallucination by over 10 points, and attaining top spatial grounding IoU, thereby setting a new standard for causally grounded medical reasoning.