探索人工智能在重症监护室脓毒症管理中的潜力。

IF 1.8 Q3 CRITICAL CARE MEDICINE
Critical Care Research and Practice Pub Date : 2025-08-28 eCollection Date: 2025-01-01 DOI:10.1155/ccrp/9031137
Ali Riahi, Mohammad Sepehr Yazdani, Reza Eshraghi, Motahare Karimi Houyeh, Ashkan Bahrami, Sara Khoshdooz, Mahshid Amini, Ehsan Behzadi, Amirreza Khalaji, Seyed Masoud Moeini Taba, Seyed Mohammad Reza Hashemian
{"title":"探索人工智能在重症监护室脓毒症管理中的潜力。","authors":"Ali Riahi, Mohammad Sepehr Yazdani, Reza Eshraghi, Motahare Karimi Houyeh, Ashkan Bahrami, Sara Khoshdooz, Mahshid Amini, Ehsan Behzadi, Amirreza Khalaji, Seyed Masoud Moeini Taba, Seyed Mohammad Reza Hashemian","doi":"10.1155/ccrp/9031137","DOIUrl":null,"url":null,"abstract":"<p><p>Sepsis remains one of the leading causes of morbidity and mortality worldwide, particularly among critically ill patients in intensive care units (ICUs). Traditional diagnostic approaches, such as the Sequential Organ Failure Assessment (SOFA) and systemic inflammatory response syndrome (SIRS) criteria, often detect sepsis after significant organ dysfunction has occurred, limiting the potential for early intervention. In this study, we reviewed how artificial intelligence (AI)-driven methodologies, including machine learning (ML), deep learning (DL), and natural language processing (NLP), can aid physicians. AI, in this case, particularly ML, processes massive amounts of real-time clinical data, vital signs, lab results, and patient history and can detect subtle patterns and predict sepsis earlier than traditional methods like SOFA or SIRS, which often lag behind after the presentation of the sequela. Models like random forest, XGBoost, and neural networks achieve high accuracy and area under the receiver operating characteristic curve (AUROC) scores (0.8-0.99) in ICU and emergency settings, enabling timely intervention by distinguishing sepsis from similar conditions despite the lack of perfect biomarkers. In practice, however, there are several potential pitfalls. Algorithmic bias due to nonrepresentative data, data fragmentation, lack of validation, and explainability issues are current barriers in developed models. Future research should address these limitations and develop more sophisticated models.</p>","PeriodicalId":46583,"journal":{"name":"Critical Care Research and Practice","volume":"2025 ","pages":"9031137"},"PeriodicalIF":1.8000,"publicationDate":"2025-08-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12411037/pdf/","citationCount":"0","resultStr":"{\"title\":\"Exploring the Potentials of Artificial Intelligence in Sepsis Management in the Intensive Care Unit.\",\"authors\":\"Ali Riahi, Mohammad Sepehr Yazdani, Reza Eshraghi, Motahare Karimi Houyeh, Ashkan Bahrami, Sara Khoshdooz, Mahshid Amini, Ehsan Behzadi, Amirreza Khalaji, Seyed Masoud Moeini Taba, Seyed Mohammad Reza Hashemian\",\"doi\":\"10.1155/ccrp/9031137\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><p>Sepsis remains one of the leading causes of morbidity and mortality worldwide, particularly among critically ill patients in intensive care units (ICUs). Traditional diagnostic approaches, such as the Sequential Organ Failure Assessment (SOFA) and systemic inflammatory response syndrome (SIRS) criteria, often detect sepsis after significant organ dysfunction has occurred, limiting the potential for early intervention. In this study, we reviewed how artificial intelligence (AI)-driven methodologies, including machine learning (ML), deep learning (DL), and natural language processing (NLP), can aid physicians. AI, in this case, particularly ML, processes massive amounts of real-time clinical data, vital signs, lab results, and patient history and can detect subtle patterns and predict sepsis earlier than traditional methods like SOFA or SIRS, which often lag behind after the presentation of the sequela. Models like random forest, XGBoost, and neural networks achieve high accuracy and area under the receiver operating characteristic curve (AUROC) scores (0.8-0.99) in ICU and emergency settings, enabling timely intervention by distinguishing sepsis from similar conditions despite the lack of perfect biomarkers. In practice, however, there are several potential pitfalls. Algorithmic bias due to nonrepresentative data, data fragmentation, lack of validation, and explainability issues are current barriers in developed models. Future research should address these limitations and develop more sophisticated models.</p>\",\"PeriodicalId\":46583,\"journal\":{\"name\":\"Critical Care Research and Practice\",\"volume\":\"2025 \",\"pages\":\"9031137\"},\"PeriodicalIF\":1.8000,\"publicationDate\":\"2025-08-28\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12411037/pdf/\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Critical Care Research and Practice\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1155/ccrp/9031137\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"2025/1/1 0:00:00\",\"PubModel\":\"eCollection\",\"JCR\":\"Q3\",\"JCRName\":\"CRITICAL CARE MEDICINE\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Critical Care Research and Practice","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1155/ccrp/9031137","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2025/1/1 0:00:00","PubModel":"eCollection","JCR":"Q3","JCRName":"CRITICAL CARE MEDICINE","Score":null,"Total":0}
引用次数: 0

摘要

脓毒症仍然是世界范围内发病和死亡的主要原因之一,特别是在重症监护病房(icu)的重症患者中。传统的诊断方法,如顺序器官衰竭评估(SOFA)和系统性炎症反应综合征(SIRS)标准,通常在发生重大器官功能障碍后检测败血症,限制了早期干预的潜力。在这项研究中,我们回顾了人工智能(AI)驱动的方法,包括机器学习(ML)、深度学习(DL)和自然语言处理(NLP)如何帮助医生。在这种情况下,人工智能,特别是机器学习,可以处理大量的实时临床数据、生命体征、实验室结果和患者病史,并且可以比SOFA或SIRS等传统方法更早地检测到细微的模式并预测败血症,这些方法通常在后遗症出现后落后。随机森林、XGBoost和神经网络等模型在ICU和急诊环境中具有较高的准确性和受试者工作特征曲线下面积(AUROC)评分(0.8-0.99),尽管缺乏完善的生物标志物,但仍可通过将脓毒症与类似情况区分开来进行及时干预。然而,在实践中,有几个潜在的陷阱。由于非代表性数据、数据碎片、缺乏验证和可解释性问题导致的算法偏差是目前开发模型的障碍。未来的研究应该解决这些限制,并开发更复杂的模型。
本文章由计算机程序翻译,如有差异,请以英文原文为准。

Exploring the Potentials of Artificial Intelligence in Sepsis Management in the Intensive Care Unit.

Exploring the Potentials of Artificial Intelligence in Sepsis Management in the Intensive Care Unit.

Exploring the Potentials of Artificial Intelligence in Sepsis Management in the Intensive Care Unit.

Sepsis remains one of the leading causes of morbidity and mortality worldwide, particularly among critically ill patients in intensive care units (ICUs). Traditional diagnostic approaches, such as the Sequential Organ Failure Assessment (SOFA) and systemic inflammatory response syndrome (SIRS) criteria, often detect sepsis after significant organ dysfunction has occurred, limiting the potential for early intervention. In this study, we reviewed how artificial intelligence (AI)-driven methodologies, including machine learning (ML), deep learning (DL), and natural language processing (NLP), can aid physicians. AI, in this case, particularly ML, processes massive amounts of real-time clinical data, vital signs, lab results, and patient history and can detect subtle patterns and predict sepsis earlier than traditional methods like SOFA or SIRS, which often lag behind after the presentation of the sequela. Models like random forest, XGBoost, and neural networks achieve high accuracy and area under the receiver operating characteristic curve (AUROC) scores (0.8-0.99) in ICU and emergency settings, enabling timely intervention by distinguishing sepsis from similar conditions despite the lack of perfect biomarkers. In practice, however, there are several potential pitfalls. Algorithmic bias due to nonrepresentative data, data fragmentation, lack of validation, and explainability issues are current barriers in developed models. Future research should address these limitations and develop more sophisticated models.

求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
Critical Care Research and Practice
Critical Care Research and Practice CRITICAL CARE MEDICINE-
CiteScore
3.60
自引率
0.00%
发文量
34
审稿时长
14 weeks
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信