---
title: "医疗思维链忠实性审计：正确诊断与装饰性推理的扰动检验"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.24790v1"
permalink: "https://feed.astralor.com/items/itm_64b771ff9237edf1bcebd8dc"
published: "2026-08-25T16:37:39.000Z"
score: 78
channel: "ai"
tags: ["医疗AI","可解释性","大语言模型","思维链"]
---

# 医疗思维链忠实性审计：正确诊断与装饰性推理的扰动检验

## AI 点评

> 这篇论文最值得看的地方是它用极其朴素的手段戳穿了一个行业幻觉：模型生成的思维链可能只是漂亮的文档，而不是真正的推理过程。对医疗AI产品经理、临床决策支持系统开发者和监管审计人员来说，CDR这个指标值得直接拿去做内部模型的例行体检。

## 速览

医学专家常将链式思维（CoT）输出视为模型临床推理的证据，但该论文用30种临床相关扰动算子对思维链和问题同时进行编辑，并在14个大模型、4个医学问答基准上开展审计。结果显示：对临床意义明确的破坏性编辑，模型的链-解耦率（CDR）高达72.9%，即思维链未感知编辑而答案也未翻转；破坏思维链不影响准确率，移除CoT提示同样不降低性能。两名认证临床医生复核197个扰动问题，98.5%的金标准仍可辩护。这表明医疗CoT更多是装饰性文档而非忠实推理，CDR可作为通用度量工具。

## 关键点

- 提出30种临床扰动算子，同时编辑思维链与问题，引入链更新×答案翻转联合分析。
- 14个LLM在4个医学QA基准上CDR均值达72.9%，破坏思维链不影响准确率，去除CoT也不降低性能。
- 两名认证临床医生复核197个扰动问题，98.5%的金标准仍可辩护。
- 提示医疗CoT可能只是装饰性推理，CDR可成为衡量医疗CoT忠实性的通用标尺。

## 全文

Clinicians read chain-of-thought (CoT) rationales as evidence of medical reasoning, but whether the visible chain plays that role is rarely tested. General-domain CoT-faithfulness probes ignore clinical cost, and medical LLM evaluations treat the chain as a black box. We close this gap with a medical perturbation audit: a 30-operator battery edits both the chain and the question with clinically motivated operators (severity reversal, negation flip, demographic swap, evidence ablation), paired with a chain-update times answer-flip joint analysis that classifies each model by its failure mode. Applied to 14 LLMs on four medical QA benchmarks, three independent tests converge: the Chain-Decoupling Rate (CDR; chain does not register the edit and the answer does not flip) is 72.9% panel-wide on clinically meaningful destructive edits, chain corruption leaves accuracy unchanged, and removing CoT prompting does not reduce accuracy. Two board-certified clinicians re-annotate N=197 perturbed questions; 98.5% leave the gold defensible. The pattern holds across medical and reasoning fine-tuning and scale; on the closed-source tier, where the chain text is unavailable, the answer-side signals are consistent with the same decoupling. Our framework and CDR provide a reusable yardstick for auditing whether medical CoT is faithful or merely documentation.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_64b771ff9237edf1bcebd8dc