ACL Findings2026
DiagnosisArena: Benchmarking Diagnostic Reasoning for Large Language Models
Yakun Zhu, Zhongzhen Huang, Linjie Mu, Yutong Huang, Wei Nie, Jiaji Liu, Shaoting Zhang, Pengfei Liu, Xiaofan Zhang
摘要
The emergence of groundbreaking large language models capable of performing complex reasoning tasks holds significant promise for addressing various scientific challenges, including those arising in complex clinical scenarios. To enable their safe and effective deployment in real-world healthcare settings, it is urgently necessary to benchmark the diagnostic capabilities of current models systematically. Given the limitations of existing medical benchmarks in evaluating advanced diagnostic reasoning, we present DiagnosisArena, a comprehensive and challenging benchmark designed to rigorously assess professional-level diagnostic competence. DiagnosisArena consists of 1,113 pairs of segmented patient cases and corresponding diagnoses, spanning 28 medical specialties, deriving from clinical case reports published in 10 top-tier medical journals. The benchmark is developed through a meticulous construction pipeline, involving multiple rounds of screening and review by both AI systems and human experts, with thorough checks conducted to prevent data leakage. Our study reveals that even the most advanced reasoning models, o3-mini, o1, and DeepSeek-R1, achieve only 45.82%, 31.09%, and 17.79% accuracy, respectively. This finding highlights a significant generalization bottleneck in current large language models when faced with clinical diagnostic reasoning challenges. Through Di-agnosisArena, we aim to drive further advancements in AI's diagnostic reasoning capabilities, enabling more effective solutions for real-world clinical diagnostic challenges. We openly share the benchmark and evaluation tools for further research and development 1 .
问 Lune
几分钟读透这篇 Paper,不用再花一下午。
Lune 能读完全文,也能读引用它的顶会 Paper,讲清它的结论、证据和局限,每一点都注明出自哪篇 Paper。不用硬啃,也不靠猜。
已有 6 篇顶会 Paper 引用了它。别等审稿人问起,先弄清它们发现了什么。
脉络
引用它的顶会 Paper6
- Real-Time Aligned Reward Model beyond SemanticsZixuan Huang, Xin Xia, Yuxi Ren, Jianbin Zheng 等ICML 2026 · 18 次引用
- Beyond Classification Accuracy: Neural-MedBench and the Need for Deeper Reasoning BenchmarksMiao Jing, Mengting Jia, Junling Lin, Zhongxia Shen 等ICLR 2026 · 4 次引用
- Graph of States: Solving Abductive Tasks with Large Language ModelsYu Luo, Rongchen Gao, Lu Teng, Xidao Wen 等ICML 2026 · 2 次引用
- "Do I Trust the AI?" Towards Trustworthy AI-Assisted Diagnosis: Understanding User Perception in LLM-Supported Clinical ReasoningYuansong Xu, Yichao Zhu, Haokai Wang, Yuchen Wu 等CHI 2026 · 1 次引用
- Salus: Strategic Diagnostic Testing for Complex Diagnosis via Multi-Agent Reinforcement LearningShuohao Gao, Xuanzhong Chen, Lingxiao Luo, Zilin Ding 等ICML 2026
它引用的 Paper4
- Measuring Massive Multitask Language UnderstandingDan Hendrycks, Collin Burns, Steven Basart, Andy Zou 等ICLR 2021 · 7,905 次引用
- Let's Verify Step by StepHunter Lightman, Vineet Kosaraju, Yuri Burda, Harrison Edwards 等ICLR 2024 · 3,045 次引用
- MedXpertQA: Benchmarking Expert-Level Medical Reasoning and UnderstandingYuxin Zuo, Shang Qu, Yifei Li, Zhang-Ren Chen 等ICML 2025
- Math-Shepherd: Verify and Reinforce LLMs Step-by-step without Human AnnotationsPeiyi Wang, Lei Li, Zhihong Shao, Runxin Xu 等ACL 2024
相关 Paper
- Benchmarking LLMs on Authentic Cases from Medical JournalsWanlong Liu, Junying Chen, Yunjin Yang, Prayag Tiwari 等ACL Findings 2026 · 1 次引用
- LLMEval-Med: A Real-world Clinical Benchmark for Medical LLMs with Physician ValidationMing Zhang, Yujiong Shen, Zelin Li, Huayu Sha 等EMNLP Findings 2025 · 6 次引用
- MedBench: A Large-Scale Chinese Benchmark for Evaluating Medical Large Language ModelsYan Cai, Linlin Wang, Ye Wang, Gerard de Melo 等AAAI 2024 · 42 次引用
- Omni-MATH: A Universal Olympiad Level Mathematic Benchmark for Large Language ModelsBofei Gao, Feifan Song, Zhe Yang, Zefan Cai 等ICLR 2025 · 3 次引用
- MentalSeek-Dx: Towards Progressive Hypothetico-Deductive Reasoning for Real-world Psychiatric DiagnosisXiao Sun, Yuming Yang, Xinyi Jiang, Yu Tian 等ACL 2026 · 1 次引用
