Search research articles

关于 JoVE

概览领导团队博客 JoVE 帮助中心

作者

出版流程编辑委员会范围与政策同行评审常见问题投稿

图书馆员

用户评价订阅访问资源图书馆顾问委员会常见问题

研究

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments 存档

教育

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual 教师资源中心教师网站

使用条款与条件

相关概念视频

您也可能阅读

相关文章

通过共同作者、期刊和引用图与本文相关的文章。

排序

Same author

WHO 2022 Renal Cell Tumor Classification: Imaging Features and Clinical Implications.

Radiographics : a review publication of the Radiological Society of North America, Inc·2026

Same author

Imaging Spectrum of Childhood Interstitial Lung Diseases: Focus on Disorders Not Specific to Infancy.

Radiographics : a review publication of the Radiological Society of North America, Inc·2026

Same author

Pediatric Rheumatic Disorders Revisited: Integrating Imaging and Pathophysiologic Insights across the Autoinflammatory-Autoimmune Continuum.

Radiographics : a review publication of the Radiological Society of North America, Inc·2026

Same author

Estimation of Lung Volume Superimposed on the Mediastinum and Diaphragm on Chest Radiographs Using Upright Multidetector Row CT.

Radiology. Cardiothoracic imaging·2026

Same author

A Case of Testicular Malakoplakia With Markedly High Signal Intensity on Fat-Suppressed T1-Weighted Images.

IJU case reports·2026

Same author

Percutaneous Cryoablation Under Local Anesthesia for Pulmonary Metastases From Colorectal Cancer: Long-Term Outcomes From a Single-Institution Retrospective Cohort.

Cancer reports (Hoboken, N.J.)·2026

Search research articles

相关实验视频

Updated: Sep 14, 2025

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

培训语言模型用于估计超声波检查优先级等待列表:算法开发和验证.

Kanato Masayoshi¹, Masahiro Hashimoto¹, Naoki Toda¹

¹Department of Radiology, School of Medicine, Keio University, 35 Shinanomachi, Shinjuku-ku, Tokyo 160-8582, Tokyo, Japan, 81 3-3353-1211 ext 62477.

|July 22, 2025

概括

人工智能语言模型可以准确地估计医疗检查请求优先级,匹配人类放射科医生的表现. 这证明了AI.

关键词:

临床信息学临床信息学健康资源卫生资源卫生资源医院信息系统. 医院信息系统.大型语言模型机器学习是机器学习.自然语言处理自然语言处理.超声波学超声波学超声波学

更多相关视频

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

Point-of-Care Ultrasound: A Review of Ultrasound Parameters for Predicting Difficult Airways

Point-of-Care Ultrasound: A Review of Ultrasound Parameters for Predicting Difficult Airways

Published on: April 7, 2023

相关实验视频

Last Updated: Sep 14, 2025

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

Published on: April 21, 2023

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Augmenting Large Language Models via Vector Embeddings to Improve Domain-Specific Responsiveness

Published on: December 6, 2024

Point-of-Care Ultrasound: A Review of Ultrasound Parameters for Predicting Difficult Airways

Point-of-Care Ultrasound: A Review of Ultrasound Parameters for Predicting Difficult Airways

Published on: April 7, 2023

科学领域:

医疗成像医学成像
人工智能的人工智能
自然语言处理自然语言处理.

背景情况:

超声波检查是必不可少的,但面临着可用性限制,需要医院预订和等候名单系统.
当前的等待列表系统往往缺乏自动优先级,需要手动审查自由形式的文本请求.
这种手动过程耗时,并且可能是患者护理中的瓶.

研究的目的:

调查使用AI语言模型来优先考虑医疗检查请求的可行性.
评估各种语言模型在处理日本医学文本中的性能,以进行优先估计.
确定应用人工智能的挑战,以优先考虑临床工作流程.

主要方法:

来自凯奥大学医院 (2020年1月-2023年3月) 的2,335次超声波检查请求的回顾性收集.
使用两种方法对四个语言模型 (JMedRoBERTa,Luke,OpenCalm,LLaMA2) 进行微调:最终层调和全层调.
使用肯德尔系数对模型性能进行评估,与放射科医生重新评估进行比较.

主要成果:

只有最后一层调整时,JMedRoBERTa 获得了最高的性能 (肯达尔系数=0.225).
在完全微调的情况下,JMedRoBERTa仍然优越 (肯德尔系数=0.254),超过其他模型和放射科医生重新评估 (肯德尔系数=0.221).
结果表明,通用模型成功地适应了日本专门的医学文本.

结论:

人工智能语言模型可以有效地估计医疗检查请求优先级,其准确性与人类放射科医生相美.
微调使通用AI模型能够处理特定领域的医学文本,显示出临床应用的前景.
未来的工作应该解决等级模糊性,多机构数据,并探索先进的语言模型.