Article

Understanding Clinical Reasoning Variability in Medical Large Language Models: A Mechanistic Interpretability Study

2026-01-27

Abstract excerpt

Medical large language models (LLMs) achieving high benchmark accuracy exhibit unexplained variability in clinical tasks, producing errors that clinicians cannot safeguard against. We evaluated clinical reasoning stability in GPT-5, MedGemma-27B-Text-IT, and OpenBioLLM-Llama3-70B using 355 systematic perturbations of physician-validated oncology cases and trained sparse autoencoders on 1 billion tokens from 50,000...

Topics

Open a Topic to create a Post that cites this publication.

Identifiers and source

Literature Corpus work: 1cc4088b-6c53-588f-afb3-5c285ecaadf6
DOI: 10.64898/2026.01.26.26344845

Open publication

Related research

Semantic proximity does not establish scientific evidence.

Click a neighbor to travelStep 1 · 12 closest

Understanding Clinical Reasoning Variability in Medical Large Language Models: A Mechanistic Interpretability StudyDOI 10.64898/2026.01.26.26344845Read

High Consistency, Limited Accuracy: Evaluating Large Language Models for Binary Medical DiagnosisSemantic proximity Read

Development and retrospective validation of SCOUT: scalable clinical oversight of large language models via uncertainty triangulationSemantic proximity Read

Grounded large language models for diagnostic prediction in real-world emergency department settingsSemantic proximity Read

On the robustness of medical term representations in locally deployable language modelsSemantic proximity Read

Evaluating a Large Reasoning Model’s Performance on Open-Ended Medical ScenariosSemantic proximity Read

Evaluating the AI Potential as a Safety Net for Diagnosis: A Novel Benchmark of Large Language Models in Correcting Diagnostic ErrorsSemantic proximity Read

Generalist Foundation Models Are Not Clinical Enough for Hospital OperationsSemantic proximity Read

Development of a RAG-based Expert LLM for Clinical Support in Radiation OncologySemantic proximity Read

Auditing frontier general-purpose large language models in biomedical tasks: reasoning gains, extraction limits, and benchmark reliabilitySemantic proximity Read

Beyond Benchmarks: Dynamic, Automatic And Systematic Red-Teaming Agents For Trustworthy Medical Language ModelsSemantic proximity Read

Learning the Phenotype of Medical HallucinationsSemantic proximity Read

Reasoning Beyond Accuracy: Expert Evaluation of Large Language Models in Diagnostic PathologySemantic proximity Read