Medical large language models are easily distracted
NYU Langone Medical Center(纽约大学朗格尼医学中心);The University of Texas at Austin(德克萨斯大学奥斯汀分校);Washington University School of Medicine in St. Louis(圣路易斯华盛顿大学医学院);New York University(纽约大学)
Krithik Vishwanath, Anton Alyakin, Daniel Alexander Alber, Jin Vivian Lee, Douglas Kondziolka, Eric Karl Oermann
Large language models (LLMs) have the potential to transform medicine, but real-world clinical scenarios contain extraneous information that can hinder performance. The rise of assistive technologies like ambient dictation, which automatically generates draft notes from live patient encounters, has the potential to introduce additional noise making it crucial to assess the ability of LLM's to filter relevant data. To investigate this, we developed MedDistractQA, a benchmark using USMLE-style questions embedded with simulated real-world distractions. Our findings show that distracting statements (polysemous words with clinical meanings used in a non-clinical context or references to unrelated health conditions) can reduce LLM accuracy by up to 17.9%. Commonly proposed solutions to improve model performance such as retrieval-augmented generation (RAG) and medical fine-tuning did not change this effect and in some cases introduced their own confounders and further degraded performance. Our findings suggest that LLMs natively lack the logical mechanisms necessary to distinguish relevant from irrelevant clinical information, posing challenges for real-world applications. MedDistractQA and our results highlights the need for robust mitigation strategies to enhance LLM resilience to extraneous information.
Negated and Misprimed Probes for Pretrained Language Models: Birds Can Talk, But Cannot Fly
在Petroni等人(2019)的基础上,我们提出了两个新的探测任务,分析存储在预训练语言模型(PLM)中的事实知识。(1)否定。我们发现,PLM没有区分否定型(“Birds cannot [MASK]”)和非否定型(“Birds can [MASK]”)完形填空题。(2)启动错误。受人类心理学中启动方法的启发,我们在完形填空题中加入了“误启动”(“谈话?鸟类可以[面具]”)。我们发现,PLM很容易被误启动分心。这些结果表明,PLM仍然有很长的路要走,充分学习人类一样的事实知识。
Medical mT5: An Open-Source Multilingual Text-to-Text LLM for The Medical Domain
HiTZ Center - Ixa, University of the Basque Country UPV/EHU(巴斯克大学 HiTZ 中心 - Ixa);Fondazione Bruno Kessler(布鲁诺·凯斯勒基金会);Cruces University Hospital(克鲁塞斯大学医院)
MedZERO: Self-Evolving Agents for Open-Ended Medical Reasoning Through Controlled Knowledge Accumulation
The Chinese University of Hong Kong(香港中文大学);Zhongshan Hospital, Fudan University(复旦大学附属中山医院);South China University of Technology(华南理工大学);Chinese Academy of Sciences(中国科学院)
Large Language Models Lack Temporal Awareness of Medical Knowledge
University of Virginia(弗吉尼亚大学);National Institutes of Health(美国国家卫生研究院);Dana-Farber Cancer Institute(达纳-法伯癌症研究所);Yale University(耶鲁大学);Weill Cornell Medicine(韦氏 Cornell 医学院)
Diagnosing Shoulder Disorders Using Multimodal Large Language Models and Consumer-Grade Cameras
Bytedance(字节跳动);Peking University(北京大学);Peking University People’s Hospital(北京大学人民医院);Mohamed bin Zayed University of Artificial Intelligence(穆罕默德·本·扎耶德人工智能大学);The Chinese University of Hong Kong(香港中文大学)
Convolutional Neural Networks for Medical Diagnosis from Admission Notes
Petuum Inc.(Petuum 公司);Department of Psychiatry, University of Pittsburgh Medical Center(匹兹堡大学医学中心精神科)
$\textbf{Objective}$开发一种自动诊断系统,该系统仅使用电子健康记录(EHR)中的文本入院信息,并通过及时且经统计学验证的决策工具协助临床医生。希望该工具可以用于减少误诊。$\textbf{Materials and Methods}$我们使用来自MIMIC-III的真实世界临床记录,这是一个免费提供的数据集,由2001年至2012年期间在贝斯以色列女执事医疗中心重症监护室住院的4万多名患者的临床数据组成。我们提出了一个卷积神经网络模型,从非结构化文本输入中学习语义特征,并自动预测初次出院诊断。$\textbf{Results}$提出的模型在10种最常见的疾病类别上实现了96.11%的总体准确性和80.48%的加权F1评分值,在加权F1评分方面显著优于4个强基线模型至少12.7%。$\textbf{Discussion}$实验结果表明,CNN模型适用于在存在复杂、噪声和非结构化临床数据的情况下支持诊断决策,同时使用比其他传统深度网络模型更少的层和参数。$\textbf{Conclusion}$我们的模型证明了从非结构化的临床笔记中代表复杂的医学有意义的特征的能力,以及对常见误诊的常见疾病的预测能力。它可以很容易地在临床环境中使用,以提供及时和统计证明的决策支持。$\textbf{Keywords}$卷积神经网络、文本分类、出院诊断预测、来自EHR的入院信息。