Clinical Reasoning

Follow this topic Ask about this topic

Clinical reasoning asks how systems can turn symptoms, records, test results, and images into diagnoses and care decisions while handling uncertainty and missing information. Current methods include retrieval from guidelines and knowledge graphs, Bayesian or symbolic reasoning, structured prompting and reinforcement learning, and agents that ask follow-up questions or coordinate specialist perspectives. Work increasingly evaluates multi-turn consultations, longitudinal EHR tasks, and multimodal cases, measuring evidence use, calibration, safety, and decision quality alongside diagnostic accuracy. A central tension is that strong exam or static-case performance does not reliably translate to consistent, evidence-grounded decisions in interactive clinical settings, making prospective validation and trustworthy process evaluation important open challenges.

Latest papers

184 papers19 papers in the last four weeks, up 6% on the four weeks before. 0.1% of all new papers.

Counting Clues: A Lightweight Probabilistic Baseline Can Match an LLM
Dec 14, 2025Furong Jia, Yuan Pu, Finn Guo +1Language Model Reasoning Evaluation Clinical QA
Duke University
Timely Clinical Diagnosis through Active Test Selection
Oct 21, 2025Silas Ruhrberg Estévez, Nicolás Astorga, Mihaela van der SchaarClinical Decision Support Clinical Decision-Making
University of Cambridge Cambridge, UK
RPRO: Ranked Preference Reinforcement Optimization for Enhancing Medical QA and Diagnostic Reasoning
Aug 31, 2025Chia-Hsuan Hsu, Jun-En Ding, Hsin-Ling Hsu +5Medical LLMs Medical QA
Department of Computer Science and Information Engineering, National Taiwan University of Science and Technology, Taipei City, Taiwan · Department of Systems Engineering, Stevens Institute of Technology, Hoboken, NJ, USA · Far Eastern Memorial Hospital, New Taipei City, Taiwan +4
EndoCogniAgent: Closed-Loop Agentic Reasoning with Self-Consistency Validation for Endoscopic Diagnosis
Aug 10, 2025Yi Tang, Kai-Ni Wang, Liang-Peng Pu +3Endoscopy Medical Imaging
School of Biological Science and Medical Engineering, Southeast University, Nanjing 211189, China · Jiangsu Key Laboratory of Biomaterials and Devices, Southeast University, Nanjing 211189, China · State Key Laboratory of Digital Medical Engineering, Southeast University, Nanjing 211189, China +2

Dec 14, 2025J/K move · Enter open · S save

Counting Clues: A Lightweight Probabilistic Baseline Can Match an LLM

Furong Jia, Yuan Pu, Finn Guo, Monica Agrawal

Duke University

Language Model Reasoning Evaluation Clinical QA Question Answering Clinical Reasoning Clinical Prediction

Open paper AskPDF arXiv

Large language models (LLMs) excel on multiple-choice clinical diagnosis benchmarks, yet it is unclear how much of this performance reflects underlying probabilistic reasoning. We study this through questions from MedQA that ask for the most likely diagnosis. We introduce Frequency-Based Probabilistic Ranker (FBPR), a lightweight method that scores options with a smoothed Naive Bayes over concept-diagnosis co-occurrence statistics from a large corpus. When co-occurrence statistics are sourced from Dolma, the pretraining corpus for OLMo, and RedPajama, an open reproduction of the LLaMA pretraining-data recipe, FBPR achieves comparable performance to the corresponding LLMs. Furthermore, direct LLM inference and FBPR exhibit complementary strengths, correctly answering largely different subsets of questions with overlap barely above chance. Combining them with a simple hybrid method yields a notable gain over either method alone. These findings show that explicit probabilistic baselines can account for a substantial portion of the benchmark performance on diagnosis questions, while serving as historically grounded reference points and complementary signals for hybridization.

Latest papers

Counting Clues: A Lightweight Probabilistic Baseline Can Match an LLM

Timely Clinical Diagnosis through Active Test Selection

RPRO: Ranked Preference Reinforcement Optimization for Enhancing Medical QA and Diagnostic Reasoning

EndoCogniAgent: Closed-Loop Agentic Reasoning with Self-Consistency Validation for Endoscopic Diagnosis