人工智能中的透明度和可重复性。
Transparency and reproducibility in artificial intelligence.
机构信息
Princess Margaret Cancer Centre, University Health Network, Toronto, Ontario, Canada.
Department of Medical Biophysics, University of Toronto, Toronto, Ontario, Canada.
出版信息
Nature. 2020 Oct;586(7829):E14-E16. doi: 10.1038/s41586-020-2766-y. Epub 2020 Oct 14.
Abstract
Breakthroughs in artificial intelligence (AI) hold enormous potential as it can automate complex tasks and go even beyond human performance. In their study, McKinney et al. showed the high potential of AI for breast cancer screening. However, the lack of methods’ details and algorithm code undermines its scientific value. Here, we identify obstacles hindering transparent and reproducible AI research as faced by McKinney et al., and provide solutions to these obstacles with implications for the broader field.
摘要
人工智能（AI）的突破具有巨大的潜力，因为它可以自动化复杂的任务，甚至超越人类的表现。在他们的研究中，McKinney 等人展示了 AI 在乳腺癌筛查方面的巨大潜力。然而，缺乏方法细节和算法代码降低了其科学价值。在这里，我们确定了 McKinney 等人在进行透明和可重复的 AI 研究时所面临的障碍，并提供了解决这些障碍的方法，这些方法对更广泛的领域也具有意义。
相似文献
1
Transparency and reproducibility in artificial intelligence.人工智能中的透明度和可重复性。
Nature. 2020 Oct;586(7829):E14-E16. doi: 10.1038/s41586-020-2766-y. Epub 2020 Oct 14.
2
Reply to: Transparency and reproducibility in artificial intelligence.回复：人工智能中的透明度与可重复性
Nature. 2020 Oct;586(7829):E17-E18. doi: 10.1038/s41586-020-2767-x.
3
Artificial Intelligence and Mechanical Circulatory Support.人工智能与机械循环支持。
Heart Fail Clin. 2022 Apr;18(2):301-309. doi: 10.1016/j.hfc.2021.11.005. Epub 2022 Mar 4.
4
Machine learning and artificial intelligence research for patient benefit: 20 critical questions on transparency, replicability, ethics, and effectiveness.机器学习和人工智能研究如何造福患者：透明度、可重复性、伦理和有效性方面的 20 个关键问题。
BMJ. 2020 Mar 20;368:l6927. doi: 10.1136/bmj.l6927.
5
A roadmap to artificial intelligence (AI): Methods for designing and building AI ready data to promote fairness.人工智能（AI）路线图：设计和构建 AI 就绪数据的方法，以促进公平性。
J Biomed Inform. 2024 Jun;154:104654. doi: 10.1016/j.jbi.2024.104654. Epub 2024 May 11.
6
Ethics and artificial intelligence.伦理学与人工智能。
Rev Clin Esp (Barc). 2024 Mar;224(3):178-186. doi: 10.1016/j.rceng.2024.02.003. Epub 2024 Feb 12.
7
The inclusion of augmented intelligence in medicine: A framework for successful implementation.医学中增强型人工智能的应用：成功实施的框架。
Cell Rep Med. 2022 Jan 18;3(1):100485. doi: 10.1016/j.xcrm.2021.100485.
8
Framework and metrics for the clinical use and implementation of artificial intelligence algorithms into endoscopy practice: recommendations from the American Society for Gastrointestinal Endoscopy Artificial Intelligence Task Force.将人工智能算法临床应用及实施于内镜检查实践的框架与指标：美国胃肠内镜学会人工智能特别工作组的建议
Gastrointest Endosc. 2023 May;97(5):815-824.e1. doi: 10.1016/j.gie.2022.10.016. Epub 2023 Feb 8.
9
Lack of Transparency and Potential Bias in Artificial Intelligence Data Sets and Algorithms: A Scoping Review.人工智能数据集和算法中缺乏透明度和潜在偏见：范围综述。
JAMA Dermatol. 2021 Nov 1;157(11):1362-1369. doi: 10.1001/jamadermatol.2021.3129.
10
Artificial intelligence and machine learning for anaphylaxis algorithms.人工智能和机器学习在过敏算法中的应用。
Curr Opin Allergy Clin Immunol. 2024 Oct 1;24(5):305-312. doi: 10.1097/ACI.0000000000001015. Epub 2024 Jul 24.
引用本文的文献
1
DeepISLES: a clinically validated ischemic stroke segmentation model from the ISLES'22 challenge.DeepISLES：一个来自ISLES'22挑战赛的经过临床验证的缺血性中风分割模型。
Nat Commun. 2025 Aug 9;16(1):7357. doi: 10.1038/s41467-025-62373-x.
2
The use of machine learning models to predict progression-free survival and overall survival outcomes from waterfall plots in randomized clinical trials (MAP-OUTCOMES).利用机器学习模型从随机临床试验的瀑布图预测无进展生存期和总生存期结果（MAP - 结果）
ESMO Open. 2025 Jul 14;10(8):105509. doi: 10.1016/j.esmoop.2025.105509.
3
Big team science reveals promises and limitations of machine learning efforts to model physiological markers of affective experience.大型团队科学揭示了机器学习在对情感体验的生理指标进行建模方面的前景与局限。
R Soc Open Sci. 2025 Jun 25;12(6):241778. doi: 10.1098/rsos.241778. eCollection 2025 Jun.
4
Concordance with SPIRIT-AI guidelines in reporting of randomized controlled trial protocols investigating artificial intelligence in oncology: a systematic review.在肿瘤学中研究人工智能的随机对照试验方案报告中与SPIRIT-AI指南的一致性：一项系统评价。
Oncologist. 2025 May 8;30(5). doi: 10.1093/oncolo/oyaf112.
5
Generalizability of FDA-Approved AI-Enabled Medical Devices for Clinical Use.美国食品药品监督管理局（FDA）批准的用于临床的人工智能医疗设备的可推广性。
JAMA Netw Open. 2025 Apr 1;8(4):e258052. doi: 10.1001/jamanetworkopen.2025.8052.
6
Is regulatory science ready for artificial intelligence?监管科学是否已为人工智能做好准备？
NPJ Digit Med. 2025 Apr 10;8(1):200. doi: 10.1038/s41746-025-01596-0.
7
Radiomic-based approaches in the multi-metastatic setting: a quantitative review.多转移情况下基于影像组学的方法：定量综述。
BMC Cancer. 2025 Mar 25;25(1):538. doi: 10.1186/s12885-025-13850-5.
8
Public Disclosure of Results From Artificial Intelligence/Machine Learning Research in Health Care: Comprehensive Analysis of ClinicalTrials.gov, PubMed, and Scopus Data (2010-2023).医疗保健领域人工智能/机器学习研究结果的公开披露：对ClinicalTrials.gov、PubMed和Scopus数据的综合分析（2010 - 2023年）
J Med Internet Res. 2025 Mar 21;27:e60148. doi: 10.2196/60148.
9
"The Machine Will See You Now": A Clinician's Perspective on Artificial "Intelligence" In Clinical Care.“机器现在将为您服务”：临床医生对临床护理中人工智能的看法。
Mov Disord Clin Pract. 2025 May;12(5):588-591. doi: 10.1002/mdc3.70054. Epub 2025 Mar 20.
10
An interpretable machine learning model based on computed tomography radiomics for predicting programmed death ligand 1 expression status in gastric cancer.一种基于计算机断层扫描影像组学的可解释机器学习模型，用于预测胃癌中程序性死亡配体1的表达状态。
Cancer Imaging. 2025 Mar 12;25(1):31. doi: 10.1186/s40644-025-00855-3.
本文引用的文献
1
Text Data Augmentation for Deep Learning.用于深度学习的文本数据增强
J Big Data. 2021;8(1):101. doi: 10.1186/s40537-021-00492-0. Epub 2021 Jul 19.
2
International evaluation of an AI system for breast cancer screening.国际乳腺癌筛查人工智能系统评估。
Nature. 2020 Jan;577(7788):89-94. doi: 10.1038/s41586-019-1799-6. Epub 2020 Jan 1.
3
Assessing Radiology Research on Artificial Intelligence: A Brief Guide for Authors, Reviewers, and Readers-From the Editorial Board.评估人工智能放射学研究：给作者、审稿人和读者的简要指南——来自编辑委员会
Radiology. 2020 Mar;294(3):487-489. doi: 10.1148/radiol.2019192515. Epub 2019 Dec 31.
4
Exploration, Inference, and Prediction in Neuroscience and Biomedicine.神经科学与生物医学中的探索、推理和预测。
Trends Neurosci. 2019 Apr;42(4):251-262. doi: 10.1016/j.tins.2019.02.001. Epub 2019 Feb 23.
5
Toward unrestricted use of public genomic data.迈向公共基因组数据的无限制使用。
Science. 2019 Jan 25;363(6425):350-352. doi: 10.1126/science.aaw1280.
6
Reproducible research practices, transparency, and open access data in the biomedical literature, 2015-2017.2015-2017 年生物医学文献中的可重复性研究实践、透明度和开放获取数据。
PLoS Biol. 2018 Nov 20;16(11):e2006930. doi: 10.1371/journal.pbio.2006930. eCollection 2018 Nov.
7
Artificial intelligence faces reproducibility crisis.人工智能面临可重复性危机。
Science. 2018 Feb 16;359(6377):725-726. doi: 10.1126/science.359.6377.725.
8
A curated mammography data set for use in computer-aided detection and diagnosis research.用于计算机辅助检测和诊断研究的精选 mammography 数据集。
Sci Data. 2017 Dec 19;4:170177. doi: 10.1038/sdata.2017.177.
9
Enhancing reproducibility for computational methods.提高计算方法的可重复性。
Science. 2016 Dec 9;354(6317):1240-1241. doi: 10.1126/science.aah6168.
10
Putting oncology patients at risk.使肿瘤患者处于危险之中。
Biotechnol Healthc. 2012 Fall;9(3):17-21.
Zotero 插件