Measuring Epistemic Resilience of LLMs Under Misleading Medical Context

• URL: http://arxiv.org/abs/2606.12291v2
• Date: Mon, 15 Jun 2026 05:53:19 GMT
• Title: Measuring Epistemic Resilience of LLMs Under Misleading Medical Context
• Abstract summary: Large language models (LLMs) now reach expert-level scores on medical licensing exams.<n>We show this assumption is fragile: when misleading context is injected into questions that LLMs originally answer correctly, they abandon the correct answer.<n>We call the ability to maintain correct judgment under adversarial context epistemic resilience, and introduce MedMisBench to measure it.
• Score: 47.20527783144983
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Large language models (LLMs) now reach expert-level scores on medical licensing exams, encouraging the assumption that high scores imply safe medical judgment while patients increasingly use them for health advice. We show this assumption is fragile: when misleading context is injected into questions that LLMs originally answer correctly, they abandon the correct answer. We call the ability to maintain correct judgment under adversarial context epistemic resilience, and introduce MedMisBench to measure it. MedMisBench contains 10,932 medical question items and 48,889 misleading context-option pairs spanning medical reasoning, agentic capability, and patient-journey evaluation. Across 11 model configurations, mean accuracy falls from 71.1% on original questions to 38.0% under focused misleading context, with 51.5% attack success. The most damaging injections are formal, rule-like fabrications: authority-framed falsehoods reach 69.5% attack success and exception-poisoning claims reach 64.1%. A 14-member clinical panel from 7 countries identified serious potential harm in 38.2% of reviewed cases. MedMisBench exposes a structural blind spot in LLM evaluation in medical settings: existing benchmarks measure what models know, but not whether they preserve correct medical judgment under misleading context.

Related papers

• DementiaCare-Bench: A Modality-Validated Video Benchmark [83.92703885588868]
  Dementia affects an estimated 57 million people worldwide.<n>For most families the hardest part of care is not memory loss but the behavioral and psychological symptoms of dementia.<n>We present DementiaCare-Bench: 56 professionally produced caregiver training videos segmented into 94 clips across nine BPSD categories.
  arXiv  Detail & Related papers  (2026-09-11T14:52:49Z)
• Large language models improve physician accuracy but lead to false reliance [27.854576188002074]
  Retrieval-augmented large language models (LLMs) promise source-linked clinical support.<n> CORA investigated how source-linked assistance affects physician decision-making.<n>Findings show that source-linked LLM assistance can improve physician accuracy while introducing a grounding-dependent safety risk.
  arXiv  Detail & Related papers  (2026-08-01T18:48:51Z)
• Quantifying and Mitigating Premature Closure in Frontier LLMs [4.513932256095551]
  Premature closure is a recognized contributor to diagnostic error but remains underexamined in large language models (LLMs)<n>We evaluated five frontier LLMs across structured and open-ended medical tasks.
  arXiv  Detail & Related papers  (2026-05-14T16:02:28Z)
• PanCanBench: A Comprehensive Benchmark for Evaluating Large Language Models in Pancreatic Oncology [48.732366302949515]
  Large language models (LLMs) have achieved expert-level performance on standardized examinations, yet multiple-choice accuracy poorly reflects real-world clinical utility and safety.<n>We developed a human-in-the-loop pipeline to create expert rubrics for de-identified patient questions.<n>We evaluated 22 proprietary and open-source LLMs using an LLM-as-a-judge framework, measuring clinical completeness, factual accuracy, and web-search integration.
  arXiv  Detail & Related papers  (2026-03-02T00:50:39Z)
• MedRECT: A Medical Reasoning Benchmark for Error Correction in Clinical Texts [0.0]
  Large language models (LLMs) show increasing promise in medical applications, but their ability to detect and correct errors in clinical texts remains under-evaluated.<n>We introduce MedRECT, a cross-lingual benchmark (Japanese/English) that formulates medical error handling as three subtasks.<n>We evaluate 9 contemporary LLMs spanning proprietary, open-weight, and reasoning families.
  arXiv  Detail & Related papers  (2025-11-01T06:19:34Z)
• EchoBench: Benchmarking Sycophancy in Medical Large Vision-Language Models [82.43729208063468]
  Recent benchmarks for medical Large Vision-Language Models (LVLMs) emphasize leaderboard accuracy, overlooking reliability and safety.<n>We study sycophancy -- models' tendency to uncritically echo user-provided information.<n>We introduce EchoBench, a benchmark to systematically evaluate sycophancy in medical LVLMs.
  arXiv  Detail & Related papers  (2025-09-24T14:09:55Z)
• PeruMedQA: Benchmarking Large Language Models (LLMs) on Peruvian Medical Exams -- Dataset Construction and Evaluation [0.6899744489931012]
  AIMS: to build a dataset of questions from medical examinations taken by Peruvian physicians pursuing specialty training.<n>We curated PeruMedQA, a multiple-choice question-answering dataset containing 8,380 questions spanning 12 medical domains.<n>Medgemma-27b-text-it outperformed all other models, achieving a proportion of correct answers exceeding 90% in several instances.
  arXiv  Detail & Related papers  (2025-09-15T02:07:26Z)
• MedOmni-45°: A Safety-Performance Benchmark for Reasoning-Oriented LLMs in Medicine [69.08855631283829]
  We introduce Med Omni-45 Degrees, a benchmark designed to quantify safety-performance trade-offs under manipulative hint conditions.<n>It contains 1,804 reasoning-focused medical questions across six specialties and three task types, including 500 from MedMCQA.<n>Results show a consistent safety-performance trade-off, with no model surpassing the diagonal.
  arXiv  Detail & Related papers  (2025-08-22T08:38:16Z)
• Medical Hallucinations in Foundation Models and Their Impact on Healthcare [71.15392179084428]
  Hallucinations in foundation models arise from autoregressive training objectives.<n>Top-performing models exceeded 97% accuracy when augmented with chain-of-thought prompting.
  arXiv  Detail & Related papers  (2025-02-26T02:30:44Z)
• CARES: A Comprehensive Benchmark of Trustworthiness in Medical Vision Language Models [92.04812189642418]
  We introduce CARES and aim to evaluate the Trustworthiness of Med-LVLMs across the medical domain. We assess the trustworthiness of Med-LVLMs across five dimensions, including trustfulness, fairness, safety, privacy, and robustness.
  arXiv  Detail & Related papers  (2024-06-10T04:07:09Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.