Large-scale observational study of AI-based patient and surgical material verification system in ophthalmology: real-world evaluation in 37 529 cases.

IF 5.6 1区 医学 Q1 HEALTH CARE SCIENCES & SERVICES
Hitoshi Tabuchi, Naofumi Ishitobi, Hodaka Deguchi, Yuta Nakaniida, Hayato Tanaka, Masahiro Akada, Mao Tanabe
{"title":"Large-scale observational study of AI-based patient and surgical material verification system in ophthalmology: real-world evaluation in 37 529 cases.","authors":"Hitoshi Tabuchi, Naofumi Ishitobi, Hodaka Deguchi, Yuta Nakaniida, Hayato Tanaka, Masahiro Akada, Mao Tanabe","doi":"10.1136/bmjqs-2024-018018","DOIUrl":null,"url":null,"abstract":"<p><strong>Background: </strong>Surgical errors in ophthalmology can have devastating consequences. We developed an artificial intelligence (AI)-based surgical safety system to prevent errors in patient identification, surgical laterality and intraocular lens (IOL) selection. This study aimed to evaluate its effectiveness in real-world ophthalmic surgical settings.</p><p><strong>Methods: </strong>In this retrospective observational before-and-after implementation study, we analysed 37 529 ophthalmic surgeries (18 767 pre-implementation, 18 762 post implementation) performed at Tsukazaki Hospital, Japan, between 1 March 2019 and 31 March 2024. The AI system, integrated with the WHO surgical safety checklist, was implemented for patient identification, surgical laterality verification and IOL authentication.</p><p><strong>Results: </strong>Post implementation, five medical errors (0.027%) occurred, with four in non-authenticated cases (where the AI system was not fully implemented or properly used), compared with one (0.0053%) pre-implementation (p=0.125). Of the four non-authenticated errors, two were laterality errors during the initial implementation period and two were IOL implantation errors involving unlearned IOLs (7.3% of cases) due to delayed AI updates. The AI system identified 30 near misses (0.16%) post implementation, vs 9 (0.048%) pre-implementation (p=0.00067), surgical laterality errors/near misses occurred at 0.039% (7/18 762) and IOL recognition at 0.29% (28/9713). The system achieved>99% implementation after 3 months. Authentication performance metrics showed high efficiency: facial recognition (1.13 attempts, 11.8 s), surgical laterality (1.05 attempts, 3.10 s) and IOL recognition (1.15 attempts, 8.57 s). Cost-benefit analysis revealed potential benefits ranging from US$181 946.94 to US$2 769 129.12 in conservative and intermediate scenarios, respectively.</p><p><strong>Conclusions: </strong>The AI-based surgical safety system significantly increased near miss detection and showed potential economic benefits. However, errors in non-authenticated cases underscore the importance of consistent system use and integration with existing safety protocols. These findings emphasise that while AI can enhance surgical safety, its effectiveness depends on proper implementation and continuous refinement.</p>","PeriodicalId":9077,"journal":{"name":"BMJ Quality & Safety","volume":" ","pages":""},"PeriodicalIF":5.6000,"publicationDate":"2024-11-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"BMJ Quality & Safety","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1136/bmjqs-2024-018018","RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"HEALTH CARE SCIENCES & SERVICES","Score":null,"Total":0}
引用次数: 0

Abstract

Background: Surgical errors in ophthalmology can have devastating consequences. We developed an artificial intelligence (AI)-based surgical safety system to prevent errors in patient identification, surgical laterality and intraocular lens (IOL) selection. This study aimed to evaluate its effectiveness in real-world ophthalmic surgical settings.

Methods: In this retrospective observational before-and-after implementation study, we analysed 37 529 ophthalmic surgeries (18 767 pre-implementation, 18 762 post implementation) performed at Tsukazaki Hospital, Japan, between 1 March 2019 and 31 March 2024. The AI system, integrated with the WHO surgical safety checklist, was implemented for patient identification, surgical laterality verification and IOL authentication.

Results: Post implementation, five medical errors (0.027%) occurred, with four in non-authenticated cases (where the AI system was not fully implemented or properly used), compared with one (0.0053%) pre-implementation (p=0.125). Of the four non-authenticated errors, two were laterality errors during the initial implementation period and two were IOL implantation errors involving unlearned IOLs (7.3% of cases) due to delayed AI updates. The AI system identified 30 near misses (0.16%) post implementation, vs 9 (0.048%) pre-implementation (p=0.00067), surgical laterality errors/near misses occurred at 0.039% (7/18 762) and IOL recognition at 0.29% (28/9713). The system achieved>99% implementation after 3 months. Authentication performance metrics showed high efficiency: facial recognition (1.13 attempts, 11.8 s), surgical laterality (1.05 attempts, 3.10 s) and IOL recognition (1.15 attempts, 8.57 s). Cost-benefit analysis revealed potential benefits ranging from US$181 946.94 to US$2 769 129.12 in conservative and intermediate scenarios, respectively.

Conclusions: The AI-based surgical safety system significantly increased near miss detection and showed potential economic benefits. However, errors in non-authenticated cases underscore the importance of consistent system use and integration with existing safety protocols. These findings emphasise that while AI can enhance surgical safety, its effectiveness depends on proper implementation and continuous refinement.

基于人工智能的眼科患者和手术材料验证系统的大规模观察研究:37 529例的真实世界评估。
背景:眼科手术失误会造成毁灭性的后果。我们开发了一种基于人工智能(AI)的手术安全系统,以防止患者识别、手术侧边和人工晶状体(IOL)选择方面的错误。本研究旨在评估其在现实世界眼科手术环境中的有效性。方法:在这项实施前后的回顾性观察研究中,我们分析了2019年3月1日至2024年3月31日期间在日本Tsukazaki医院进行的37529例眼科手术(实施前18767例,实施后18762例)。人工智能系统与世卫组织手术安全清单相结合,用于患者识别、手术侧边性验证和人工晶状体认证。结果:实施后,发生了5起医疗差错(0.027%),其中4起为未经认证的病例(人工智能系统未得到充分实施或使用不当),而实施前为1起(0.0053%)(p=0.125)。在4例未经验证的错误中,2例是最初实施期间的侧侧错误,2例是人工智能更新延迟导致的人工晶状体植入错误,涉及未学习的人工晶状体(7.3%)。人工智能系统在实施后识别出30例(0.16%)近距离失误,而在实施前识别出9例(0.048%)(p=0.00067),手术侧偏错误/近距离失误发生率为0.039% (7/18 762),IOL识别率为0.29%(28/9713)。经过3个月的测试,系统实现了99%的实施率。认证性能指标显示,面部识别(1.13次,11.8 s)、手术侧侧识别(1.05次,3.10 s)和人工晶状体识别(1.15次,8.57 s)具有较高的效率。成本效益分析显示,在保守和中等情景下,潜在效益分别为18946.94美元至2769 129.12美元。结论:基于人工智能的手术安全系统显著提高了近漏检率,具有潜在的经济效益。然而,在未经认证的情况下,错误强调了一致的系统使用和与现有安全协议集成的重要性。这些发现强调,虽然人工智能可以提高手术安全性,但其有效性取决于正确的实施和不断的改进。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
BMJ Quality & Safety
BMJ Quality & Safety HEALTH CARE SCIENCES & SERVICES-
CiteScore
9.80
自引率
7.40%
发文量
104
审稿时长
4-8 weeks
期刊介绍: BMJ Quality & Safety (previously Quality & Safety in Health Care) is an international peer review publication providing research, opinions, debates and reviews for academics, clinicians and healthcare managers focused on the quality and safety of health care and the science of improvement. The journal receives approximately 1000 manuscripts a year and has an acceptance rate for original research of 12%. Time from submission to first decision averages 22 days and accepted articles are typically published online within 20 days. Its current impact factor is 3.281.
×
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:481959085
Book学术官方微信