6arXiv cs.CL (Computation and Language)·
ThReadMed-QA benchmark reveals frontier LLMs degrade sharply on medical misconception correction across multi-turn conversations
Researchers introduce ThReadMed-QA, a multi-turn medical dialogue dataset of 2,437 conversation threads (8,204 QA pairs) derived from real AskDocs patient interactions, designed to evaluate LLM ability to detect and correct embedded misconceptions over multiple turns. Evaluation of five LLMs using an LLM-as-a-Judge rubric finds that even frontier models like GPT-5 and Claude Haiku drop from ~85% misconception correction on initial questions to ~50% within two follow-up turns. Oracle analysis attributes much of this degradation to error propagation from prior model outputs rather than context length alone. The findings highlight a significant safety gap in current evaluation frameworks, which do not capture multi-turn conversational dynamics in medical settings.
Related guides (3)
Related events (8)
MedMisBench: LLMs show fragile epistemic resilience under misleading medical context
Researchers introduce MedMisBench, a benchmark of 10,932 medical questions paired with 48,889 misleading context injections, to measure whether LLMs maintain correct medical judgment under adversarial pressure. Across 11 model configurations, mean accuracy drops from 71.1% to 38.0% when misleading context is injected, with authority-framed falsehoods achieving 69.5% attack success. A 14-member international clinical panel flagged serious potential harm in 38.2% of reviewed cases. The work argues that existing medical benchmarks measure knowledge but not robustness to manipulation, exposing a structural gap in LLM safety evaluation for healthcare.
Systematic evaluation of LLM prompt sensitivity in healthcare settings reveals safety risks
Researchers conduct a sensitivity analysis of both general-purpose and medical-specific LLMs using the MedMCQA benchmark, testing robustness to lexical and syntactic prompt perturbations. The study finds that even minor phrasing changes can alter clinical advice, and adversarial prompts can produce dangerous outputs such as incorrect dosages or omitted critical findings. Both general-purpose models (GPT-3.5, Llama 3) and domain-specific models (ClinicalBERT, BioLlama3, BioBERT) exhibit this fragility, with syntactic reordering and misleading contextual cues proving more destabilizing than simple paraphrasing.
MedQADE benchmark reveals LLM evaluators match physician agreement scores but lack clinical caution and show lineage bias
Researchers introduce MedQADE, a standardized open-response clinical benchmark for German comprising 3,800 items annotated by ten physicians and nine LLM evaluators. The top LLM evaluator (Gemini 3 Flash) reached statistical alignment near the physician inter-rater ceiling (κ=0.694 vs. κ=0.709), but automated evaluators showed near-zero clinical metacognition: unlike physicians, they never abstained regardless of item difficulty. The study also documents systematic lineage-dependent scoring bias, where models preferentially rate architectural siblings more favorably, independent of language.
Clinically grounded privacy evaluation framework reveals high memorization risk in medical LMs
Researchers introduce a tiered adversarial framework for evaluating privacy leakage in medical language models, moving beyond simple training-text recovery to realistic clinical threat models. Applied to an LM pretrained on 378k clinical notes, the framework finds that routine encounter metadata (name, DOB, provider, visit date) elicits high verbatim memorization and sensitive-diagnosis recovery (AUROC 0.91 for abortion, 0.81 for HIV). The study also finds that exact-match memorization overstates disclosure risk because 36% of memorized tokens reflect templated documentation. The work provides a practical contextual privacy evaluation methodology for medical LMs trained on longitudinal patient data.
New Polish medical exam benchmark reveals MCQA overestimates LLM clinical competence
Researchers introduce an expanded Polish medical exam benchmark with over 15,000 new questions, two new domains, and four structural modifications designed to reduce multiple-choice artifacts and better test reasoning. Evaluating 21 LLMs under the harder setup, the best-performing model (Qwen3.5-122B) drops 28-31 percentage points compared to standard MCQA scores. The findings suggest standard MCQA benchmarks do not reliably reflect true medical competence, even when data contamination is low. The benchmark is publicly released to support further research.
KG grounding helps LLMs only for out-of-training knowledge: controlled clinical QA study
A new arXiv paper investigates when knowledge-graph (KG) grounding improves LLM performance on clinical question answering, finding that structured KG retrieval over the public biomedical graph PrimeKG provides no meaningful improvement on MedQA (all deltas ≤3.4) because the relevant facts are already in the model's training data. On synthetic counterfactual and hybrid benchmarks containing genuinely novel facts, the same pipeline lifts out-of-training accuracy from chance to ~100%. The paper also reproduces and partially corrects a recent Nature Medicine study on frontier LLMs vs. clinical RAG tools, flagging a score-deflating grader bug and clarifying that the reported ~88 HealthBench score reflects the Consensus variant, not full HealthBench (~46-47). The core finding — that RAG/KG grounding pays off only when the decisive fact is outside the model's training distribution — has direct implications for when retrieval augmentation is worth deploying.
Training framework reduces calibration error 60%+ in Medical VQA multimodal LLMs
A new arXiv preprint proposes a finetuning framework to improve verbalized uncertainty calibration in multimodal LLMs applied to Medical Visual Question Answering. The composite loss function combines Brier-style calibration, anchor regularization, contrastive image-text alignment, and KL-based stabilization, evaluated on MedGemma 4B IT and Qwen2-VL 7B Instruct across three medical VQA benchmarks. The method reduces calibration error by 60% or more and improves discrimination by 26% or more while preserving predictive accuracy, outperforming prompting-, sampling-, and training-based baselines.
Study finds LLM health chatbot triage outcomes vary significantly with patient communication style
Researchers analyzed 2,053 real patient-chatbot conversations and found wide variation in communication patterns and emotional expression, revealing that idealized patient simulations used in chatbot development are inadequate. They built a patient simulator modeling clinical content, emotional state, conversational strategy, and communication style, achieving near-indistinguishability from real conversations (human graders 55% accurate). Evaluating four LLMs across 1,164 clinician-graded cases with five patient personae, the study found that communication style significantly alters triage urgency assessments. The authors warn that systems optimized for cooperative, articulate users risk underperforming and amplifying health disparities in real-world deployment.