Development and External Validation of a Multidimensional Deep Learning Model to Dynamically Predict Kidney Outcomes in IgA Nephropathy.

IF 8.5 1区 医学 Q1 UROLOGY & NEPHROLOGY
Tingyu Chen, Tiange Chen, Wenjie Xu, Shaoshan Liang, Feng Xu, Dandan Liang, Xiang Li, Caihong Zeng, Guotong Xie, Zhihong Liu
{"title":"Development and External Validation of a Multidimensional Deep Learning Model to Dynamically Predict Kidney Outcomes in IgA Nephropathy.","authors":"Tingyu Chen, Tiange Chen, Wenjie Xu, Shaoshan Liang, Feng Xu, Dandan Liang, Xiang Li, Caihong Zeng, Guotong Xie, Zhihong Liu","doi":"10.2215/CJN.0000000000000471","DOIUrl":null,"url":null,"abstract":"","PeriodicalId":50681,"journal":{"name":"Clinical Journal of the American Society of Nephrology","volume":null,"pages":null},"PeriodicalIF":8.5000,"publicationDate":"2024-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11254022/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Clinical Journal of the American Society of Nephrology","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.2215/CJN.0000000000000471","RegionNum":1,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2024/5/10 0:00:00","PubModel":"Epub","JCR":"Q1","JCRName":"UROLOGY & NEPHROLOGY","Score":null,"Total":0}
引用次数: 0
开发多维深度学习模型并进行外部验证,以动态预测 IgA 肾病的肾脏预后。
背景:准确预测 IgA 肾病的肾脏预后对临床决策至关重要。以往的研究中对纵向数据的使用不足,限制了预测模型的准确性和可解释性,因为它未能反映 IgA 肾病的慢性性质。本研究旨在利用全面的纵向数据建立一个多变量动态深度学习模型,用于预测IgA肾病的肾脏预后:在这项回顾性队列研究中,18 家肾脏中心共收治了 2,056 名 IgA 肾病患者,采用滑动窗口法收集了 28,317 个数据点。其中,单个中心的 15,462 个窗口被随机分配到训练集(80%)和验证集(20%),而 18 个肾脏中心的 8797 个窗口被分配到独立的测试集。利用深度学习模型--可解释多变量长短期记忆(IMV-LSTM),根据活检时测量的时变变量和随访期间测量的时变变量预测肾脏结局(肾衰竭或肾功能下降50%)。使用卡普兰-梅耶尔分析和C统计量评估风险表现。还进行了轨迹分析,以评估随访期间临床变量的各种趋势:该模型在测试集上的 C 统计量(0.93;95% CI,0.92-0.95)高于我们在之前的研究中仅使用基线信息开发的 XGBoost 预测模型(C 统计量,0.84;95% CI,0.80-0.88)。Kaplan-Meier 分析表明,完整模型预测风险较低的组比预测风险较高的组存活时间更长。时间变量的重要性得分高于时间不变变量。在时间变量中,最近的测量结果显示出更高的重要性得分。进一步的解释显示,时间变量的某些轨迹组,如血清肌酐和尿蛋白,与不良后果风险升高有关:在IgA肾病中,深度学习模型可用于根据纵向数据准确、动态地预测肾脏预后,时间变量显示出预测肾脏预后的强大能力。
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
CiteScore
12.20
自引率
3.10%
发文量
514
审稿时长
3-6 weeks
期刊介绍: The Clinical Journal of the American Society of Nephrology strives to establish itself as the foremost authority in communicating and influencing advances in clinical nephrology by (1) swiftly and effectively disseminating pivotal developments in clinical and translational research in nephrology, encompassing innovations in research methods and care delivery; (2) providing context for these advances in relation to future research directions and patient care; and (3) becoming a key voice on issues with potential implications for the clinical practice of nephrology, particularly within the United States. Original manuscript topics cover a range of areas, including Acid/Base and Electrolyte Disorders, Acute Kidney Injury and ICU Nephrology, Chronic Kidney Disease, Clinical Nephrology, Cystic Kidney Disease, Diabetes and the Kidney, Genetics, Geriatric and Palliative Nephrology, Glomerular and Tubulointerstitial Diseases, Hypertension, Maintenance Dialysis, Mineral Metabolism, Nephrolithiasis, and Transplantation.
×
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