将人工智能开发转移到眼科部署：有必要进行随机对照试验-动脉网

Moving artificial intelligence development to deployment in ophthalmology: randomised controlled trials are warranted

Nature 等信源发布 2025-02-15 04:11



可切换为仅中文







Access through your institution

通过您的机构访问

Buy or subscribe

购买或订阅

Artificial intelligence (AI) has shown promising potential in ophthalmology, evidenced by a surge in publications in the field [

人工智能（AI）在眼科领域显示出巨大的潜力，该领域出版物的激增证明了这一点[

]. However, most AI models were developed and tested in laboratory settings using retrospective datasets isolated from real-world clinical practice [

]。然而，大多数人工智能模型是在实验室环境中使用从现实世界的临床实践中分离出来的回顾性数据集开发和测试的[

]. A typical gap exists due to the variances in the laboratory and the real-world settings, including disease prevalence, concurrent diseases, population demographic characteristics, image quality, and image devices, leading to potential AI model bias, performance discrepancies, and safety issues. Nevertheless, even some AI-enabled medical devices approved by the U.S.

]。由于实验室和现实环境的差异，包括疾病流行率，并发疾病，人口统计学特征，图像质量和图像设备，存在典型的差距，导致潜在的AI模型偏差，性能差异和安全问题。尽管如此，甚至一些美国批准的人工智能医疗设备。

Food and Drug Administration (FDA) were not evaluated rigorously in real-world clinical practice [.

食品和药物管理局（FDA）在现实世界的临床实践中没有得到严格的评估[。

], which may raise concerns among regulatory bodies, clinicians, and patients [

]，这可能会引起监管机构、临床医生和患者的担忧[

In ophthalmology, real-world applications of AI models could be unreliable and underperformed based on existing evidence [

在眼科领域，根据现有证据，人工智能模型的实际应用可能不可靠且表现不佳[

]. For example, Lee et al. conducted a head-to-head real-world validation study of seven automated AI diabetic retinopathy (DR) screening systems, including FDA approval algorithms. The study revealed that one system had a sensitivity of 74.42% for detecting proliferative DR, missing nearly a quarter of advanced cases, thereby raising serious concerns about AI safety [.

]。例如，Lee等人对包括FDA批准算法在内的七种自动化AI糖尿病视网膜病变（DR）筛查系统进行了面对面的现实验证研究。该研究表明，一个系统检测增殖性DR的灵敏度为74.42%，错过了近四分之一的晚期病例，从而引起了对AI安全性的严重担忧[。

]. On the other hand, IDx-DR, one of the first FDA-approved AI systems for DR detection, demonstrated a relatively higher false-positive rate during applications than in retrospective studies [

]。另一方面，IDx DR是FDA批准的首批用于DR检测的人工智能系统之一，在应用过程中显示出比回顾性研究更高的假阳性率[

]. Such discrepancies underscore the risks of misdiagnosis and unnecessary interventions, potentially compromising patient care. In addition, observational studies reported conflicting results on the efficacy of AI models in real-world applications, with biases and variable quality limiting the consistency of evidence [.

]。这种差异强调了误诊和不必要干预的风险，可能会影响患者护理。此外，观察性研究报告了AI模型在现实世界应用中的功效相互矛盾的结果，偏倚和可变质量限制了证据的一致性[。

]. Considering the potential risks of AI in ophthalmology, rigorous testing in real-world clinical practice must align with the potential impact on patient safety.

]。考虑到人工智能在眼科的潜在风险，现实世界临床实践中的严格测试必须与对患者安全的潜在影响相一致。

This is a preview of subscription content,

这是订阅内容的预览，

access via your institution

通过您的机构访问

Access options

访问选项

Access through your institution

通过您的机构访问

Access through your institution

通过您的机构访问

Change institution

变革机构

Buy or subscribe

购买或订阅

Subscribe to this journal

订阅此日记

Receive 18 print issues and online access

接收18个打印问题和在线访问

251,40 € per year

每年251,40欧元

only 13,97 € per issue

Learn more

了解更多信息

Buy this article

购买这篇文章

Purchase on SpringerLink

在SpringerLink上购买

Instant access to full article PDF

即时访问全文PDF

Buy now

立即购买

Prices may be subject to local taxes which are calculated during checkout

价格可能需要缴纳结帐时计算的地方税

Additional access options:

其他访问选项：

Learn about institutional subscriptions

了解机构订阅

Read our FAQs

阅读我们的常见问题

Contact customer support

联系客户支持

Fig. 1

图1

References

参考文献

Ting DSW, Peng L, Varadarajan AV, Keane PA, Burlina PM, Chiang MF, et al. Deep learning in ophthalmology: The technical and clinical considerations. Prog Retin Eye Res. 2019;72:100759.

Ting DSW，Peng L，Varadarajan AV，Keane PA，Burlina PM，Chiang MF等。眼科深度学习：技术和临床考虑。Prog Retin Eye Res.2019；72:100759。

Article

文章

PubMed

Google Scholar

谷歌学者

Lam C, Wong YL, Tang Z, Hu X, Nguyen TX, Yang D, et al. Performance of artificial intelligence in detecting diabetic macular edema from fundus photography and optical coherence tomography images: a systematic review and meta-analysis. Diabetes Care. 2024;47:304–19.

Lam C，Wong YL，Tang Z，Hu X，Nguyen TX，Yang D，等。人工智能在从眼底照相和光学相干断层扫描图像检测糖尿病黄斑水肿中的表现：系统评价和荟萃分析。。2024年；47:304-19。

Article

文章

PubMed

Google Scholar

谷歌学者

Wu E, Wu K, Daneshjou R, Ouyang D, Ho DE, Zou J. How medical AI devices are evaluated: limitations and recommendations from an analysis of FDA approvals. Nat Med. 2021;27:582–4.

Wu E，Wu K，Daneshjou R，Ouyang D，Ho DE，Zou J.如何评估医疗AI设备：FDA批准分析的局限性和建议。Nat Med。2021；27:582-4。

Article

文章

CAS

中科院

PubMed

Google Scholar

谷歌学者

Rajpurkar P, Chen E, Banerjee O, Topol EJ. AI in health and medicine. Nat Med. 2022;28:31–8.

Rajpurkar P，Chen E，Banerjee O，Topol EJ。AI在健康和医学方面。Nat Med。2022；。

Article

文章

CAS

中科院

PubMed

Google Scholar

谷歌学者

Wong A, Otles E, Donnelly JP, Krumm A, McCullough J, DeTroyer-Cooley O, et al. External validation of a widely implemented proprietary sepsis prediction model in hospitalized patients. JAMA Intern Med. 2021;181:1065–70.

Wong A，Otles E，Donnelly JP，Krumm A，McCullough J，DeTroyer Cooley O等。住院患者中广泛实施的专有脓毒症预测模型的外部验证。JAMA实习生医学2021；181:1065-70年。

Article

文章

PubMed

Google Scholar

谷歌学者

DeGrave AJ, Janizek JD, Lee S-I. AI for radiographic COVID-19 detection selects shortcuts over signal. medRxiv. 2020.

DeGrave AJ，Janizek JD，Lee S-I.用于放射照相COVID-19检测的AI选择信号的快捷方式。medRxiv。2020

Lee AY, Yanagihara RT, Lee CS, Blazes M, Jung HC, Chee YE, et al. Multicenter, head-to-head, real-world validation study of seven automated artificial intelligence diabetic retinopathy screening systems. Diabetes Care. 2021;44:1168–75.

Lee AY，Yanagihara RT，Lee CS，Blazes M，Jung HC，Chee YE等。七种自动人工智能糖尿病视网膜病变筛查系统的多中心，面对面，现实世界验证研究。。；。

Article

文章

PubMed

PubMed Central

PubMed 中央

Google Scholar

谷歌学者

Abràmoff MD, Lou Y, Erginay A, Clarida W, Amelon R, Folk JC. et al. Improved Automated Detection of Diabetic Retinopathy on a Publicly Available Dataset Through Integration of Deep Learning. Invest Ophthalmol Vis Sci. 2016;57:5200-6.

Abràmoff MD，Lou Y，Erginay A，Clarida W，Amelon R，Folk JC。等人。通过整合深度学习，在公开可用的数据集上改进了糖尿病视网膜病变的自动检测。投资眼科Vis Sci。2016年；57:5200-6年。

Abramoff MD, Lavin PT, Birch M, Shah N, Folk JC. Pivotal trial of an autonomous AI-based diagnostic system for detection of diabetic retinopathy in primary care offices. NPJ Digit Med. 2018;1:39.

Abramoff MD，Lavin PT，Birch M，Shah N，Folk JC。基于人工智能的自主诊断系统在初级保健办公室检测糖尿病视网膜病变的关键试验。NPJ数字医学2018；1： 39页。

Article

文章

PubMed

PubMed Central

PubMed 中央

Google Scholar

谷歌学者

Nagendran M, Chen Y, Lovejoy CA, Gordon AC, Komorowski M, Harvey H, et al. Artificial intelligence versus clinicians: systematic review of design, reporting standards, and claims of deep learning studies. BMJ. 2020;368:m689.

Nagendran M，Chen Y，Lovejoy CA，Gordon AC，Komorowski M，Harvey H等。人工智能与临床医生：对设计，报告标准和深度学习研究主张的系统评价。英国医学杂志。；368:m689。

Article

文章

PubMed

PubMed Central

PubMed 中央

Google Scholar

谷歌学者

Youssef A, Nichol AA, Martinez-Martin N, Larson DB, Abramoff M, Wolf RM, et al. Ethical considerations in the design and conduct of clinical trials of artificial intelligence. JAMA Netw Open. 2024;7:e2432482.

Youssef A，Nichol AA，Martinez Martin N，Larson DB，Abramoff M，Wolf RM等。人工智能临床试验设计和实施中的伦理考虑。JAMA网络打开。2024年；7： e2432482。

Article

文章

PubMed

PubMed Central

公共医学中心

Google Scholar

谷歌学者

Patel MR, Balu S, Pencina MJ. Translating AI for the Clinician. JAMA. 2024.

。为临床医生翻译AI。杰玛。2024

Vasey B, Nagendran M, Campbell B, Clifton DA, Collins GS, Denaxas S, et al. Reporting guideline for the early stage clinical evaluation of decision support systems driven by artificial intelligence: DECIDE-AI. BMJ. 2022;377:e070904.

Vasey B，Nagendran M，Campbell B，Clifton DA，Collins GS，Denaxas S等。人工智能驱动的决策支持系统早期临床评估报告指南：DECIDE-AI。BMJ。2022年；377:e070904。

Article

文章

PubMed

PubMed Central

PubMed 中央

Google Scholar

谷歌学者

Hariton E, Locascio JJ. Randomised controlled trials - the gold standard for effectiveness research Study design: randomised controlled trials. Bjog-Int J Obstet Gy. 2018;125:1716.

Hariton E，Locascio JJ。随机对照试验-有效性研究设计的金标准：随机对照试验。Bjog Int J Obstet Gy。2018；125:1716年。

Article

文章

Google Scholar

谷歌学者

Han RY, Acosta JN, Shakeri Z, Ioannidis JPA, Topol EJ, Rajpurkar P. Randomised controlled trials evaluating artificial intelligence in clinical practice: a scoping review. Lancet Digit Health. 2024;6:e367–e73.

Han RY，Acosta JN，Shakeri Z，Ioannidis JPA，Topol EJ，Rajpurkar P.评估临床实践中人工智能的随机对照试验：范围界定综述。柳叶刀数字健康。2024年；6：。

Article

文章

CAS

中科院

PubMed

PubMed Central

公共医学中心

Google Scholar

谷歌学者

Liu SL, Gonder JR, Owrangi E, Klar NS, Hramiak IM, Uvarov A, et al. Effectiveness of nonmydriatic ultra-widefield retinal imaging to screen for diabetic eye disease: a randomized controlled trial (Clearsight). Diabetes Care. 2023;46:399–407.

。。2023年；46:399-407。

Article

文章

CAS

中科院

PubMed

Google Scholar

谷歌学者

Mathenge W, Whitestone N, Nkurikiye J, Patnaik JL, Piyasena P, Uwaliraye P, et al. Impact of artificial intelligence assessment of diabetic retinopathy on referral service uptake in a low-resource setting: the RAIDERS randomized trial. Ophthalmol Sci. 2022;2:100168.

Mathenge W，Whitestone N，Nkurikiye J，Patnaik JL，Piyasena P，Uwaliraye P等。人工智能评估糖尿病视网膜病变对低资源环境下转诊服务摄取的影响：RAIDERS随机试验。眼科科学。2022年；2： 100168页。

Article

文章

PubMed

PubMed Central

PubMed 中央

Google Scholar

谷歌学者

Liu XX, Rivera SC, Moher D, Calvert MJ, Denniston AK. Grp S-AC-AW. Reporting guidelines for clinical trial reports for interventions involving artificial intelligence: the CONSORT-AI extension. Nat Med. 2020;26:1364–74.

刘XX，里维拉SC，莫尔D，卡尔弗特MJ，丹尼斯顿AK。Grp S-AC-AW。涉及人工智能干预的临床试验报告报告指南：CONSORT-AI扩展。Nat Med。2020；26:1364年至74年。

Article

文章

CAS

中科院

PubMed

PubMed Central

PubMed 中央

Google Scholar

谷歌学者

Cruz Rivera S, Liu X, Chan A-W, Denniston AK, Calvert MJ, Darzi A, et al. Guidelines for clinical trial protocols for interventions involving artificial intelligence: the SPIRIT-AI extension. Nat Med. 2020;26:1351–63.

Cruz Rivera S，Liu X，Chan A-W，Denniston AK，Calvert MJ，Darzi A等。涉及人工智能干预的临床试验方案指南：SPIRIT-AI扩展。Nat Med。2020；26:1351-63。

Article

文章

CAS

中科院

PubMed

PubMed Central

PubMed 中央

Google Scholar

谷歌学者

Rajpurkar P, Lungren MP. The current and future state of ai interpretation of medical images. N. Engl J Med. 2023;388:1981–90.

Rajpurkar P，Lungren MP。医学图像人工智能解释的现状和未来。N、英国医学杂志2023；388:1981年至90年。

Article

文章

PubMed

Google Scholar

谷歌学者

Damschroder LJ, Reardon CM, Widerquist MAO, Lowery J. The updated Consolidated Framework for Implementation Research based on user feedback. Implement Sci. 2022;17:75.

Damschroder LJ，Reardon CM，Widerquist MAO，Lowery J.基于用户反馈的实施研究的更新整合框架。实施Sci。2022年；17： 75岁。

Article

文章

PubMed

PubMed Central

PubMed 中央

Google Scholar

谷歌学者

Download references

下载参考资料

Funding

资金

Health and Medical Research Fund, Hong Kong SAR (ref. 19201991).

香港特别行政区健康与医学研究基金（参考文献19201991）。

Author information

作者信息

Authors and Affiliations

作者和隶属关系

Department of Ophthalmology and Visual Sciences, The Chinese University of Hong Kong, Hong Kong SAR, China

香港中文大学眼科与视觉科学系，香港特别行政区

Shuyi Zhang, Truong X. Nguyen, Xujia Liu, Simon K. H. Szeto, Carol Y. Cheung & An Ran Ran

张淑怡，阮楚雄，刘旭佳，司徒华，张安然

Hong Kong Eye Hospital, Hong Kong SAR, China

中国香港特别行政区香港眼科医院

Simon K. H. Szeto

西蒙·K·H·塞托

Authors

作者

Shuyi Zhang

张树义

View author publications

查看作者出版物

You can also search for this author in

您也可以在中搜索此作者

PubMed

Google Scholar

谷歌学者

Truong X. Nguyen

张X.阮

View author publications

查看作者出版物

You can also search for this author in

您也可以在中搜索此作者

PubMed

Google Scholar

谷歌学者

Xujia Liu

刘 Xujia

View author publications

查看作者出版物

You can also search for this author in

您也可以在中搜索此作者

PubMed

Google Scholar

谷歌学者

Simon K. H. Szeto

西蒙·K·H·塞托

View author publications

查看作者出版物

You can also search for this author in

您也可以在中搜索此作者

PubMed

Google Scholar

谷歌学者

Carol Y. Cheung

卡罗尔·Y·张

View author publications

查看作者出版物

You can also search for this author in

您也可以在中搜索此作者

PubMed

Google Scholar

谷歌学者

An Ran Ran

View author publications

查看作者出版物

You can also search for this author in

您也可以在中搜索此作者

PubMed

Google Scholar

谷歌学者

Contributions

捐款

CYC and ARR initiated the idea. SYZ, ARR, TXN, and XJL drafted the manuscript. SKHS and CYC reviewed and refined the manuscript. All the authors have read the final version and approved the submission.

CYC和ARR提出了这个想法。SYZ，ARR，TXN和XJL起草了手稿。SKHS和CYC审查并改进了手稿。所有作者都阅读了最终版本并批准了提交。

Corresponding author

通讯作者

Correspondence to

通信对象

An Ran Ran

一个Ran Ran

Ethics declarations

道德宣言

Competing interests

相互竞争的利益

Carol Y Cheung is a member of the Eye editorial board. There are no other conflicts of interest.