Wentai Zhang, Xueyang Wu, He Wang, Ruopei Wu, Congcong Deng, Qian Xu, Xiaohai Liu, Xuexue Bai, Shuangjian Yang, Xiaoxu Li, Ming Feng, Qiang Yang, Renzhi Wang
{"title":"用于预测肢端肥大症患者术后缓解的联合学习:一项多中心研究","authors":"Wentai Zhang, Xueyang Wu, He Wang, Ruopei Wu, Congcong Deng, Qian Xu, Xiaohai Liu, Xuexue Bai, Shuangjian Yang, Xiaoxu Li, Ming Feng, Qiang Yang, Renzhi Wang","doi":"10.1016/j.wneu.2024.10.091","DOIUrl":null,"url":null,"abstract":"<p><strong>Background: </strong>Decentralized federated learning (DFL) may serve as a useful framework for machine learning (ML) tasks in multicentered studies, maximizing the use of clinical data without data sharing. We aim to propose the first workflow of DFL for ML tasks in multicentered studies, which can be as powerful as those using centralized data.</p><p><strong>Methods: </strong>A DFL workflow was developed with 4 steps: registration, local computation, model update, and inspection. A total of 598 participants with acromegaly from Peking Union Medical College Hospital, and 120 participants from Xuanwu Hospital were enrolled. The cohort from Peking Union Medical College Hospital was further split into 5 centers. Nine clinical features were incorporated into ML-based models trained based on 4 algorithms: logistic regression (LR), gradient boosted decision tree, support vector machine (SVM), and deep neural network (DNN). The area under the curve of receiver operating characteristic curves was used to evaluate the performance of the models.</p><p><strong>Results: </strong>Models trained based on DFL workflow performed better than most models in LR (P < 0.05), all models in DNN, SVM, and gradient boosted decision tree (P < 0.05). Models trained on DFL workflow performed as powerful as models trained on centralized data in LR, DNN, and SVM (P > 0.05).</p><p><strong>Conclusions: </strong>We demonstrate that the DFL workflow without data sharing should be a more appropriate method in ML tasks in multicentered studies. And the DFL workflow should be further exploited in clinical researches in other departments and it can encourage and facilitate multicentered studies.</p>","PeriodicalId":23906,"journal":{"name":"World neurosurgery","volume":" ","pages":"1036-1046"},"PeriodicalIF":1.9000,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Federated Learning for Predicting Postoperative Remission of Patients with Acromegaly: A Multicentered Study.\",\"authors\":\"Wentai Zhang, Xueyang Wu, He Wang, Ruopei Wu, Congcong Deng, Qian Xu, Xiaohai Liu, Xuexue Bai, Shuangjian Yang, Xiaoxu Li, Ming Feng, Qiang Yang, Renzhi Wang\",\"doi\":\"10.1016/j.wneu.2024.10.091\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><strong>Background: </strong>Decentralized federated learning (DFL) may serve as a useful framework for machine learning (ML) tasks in multicentered studies, maximizing the use of clinical data without data sharing. We aim to propose the first workflow of DFL for ML tasks in multicentered studies, which can be as powerful as those using centralized data.</p><p><strong>Methods: </strong>A DFL workflow was developed with 4 steps: registration, local computation, model update, and inspection. A total of 598 participants with acromegaly from Peking Union Medical College Hospital, and 120 participants from Xuanwu Hospital were enrolled. The cohort from Peking Union Medical College Hospital was further split into 5 centers. Nine clinical features were incorporated into ML-based models trained based on 4 algorithms: logistic regression (LR), gradient boosted decision tree, support vector machine (SVM), and deep neural network (DNN). The area under the curve of receiver operating characteristic curves was used to evaluate the performance of the models.</p><p><strong>Results: </strong>Models trained based on DFL workflow performed better than most models in LR (P < 0.05), all models in DNN, SVM, and gradient boosted decision tree (P < 0.05). Models trained on DFL workflow performed as powerful as models trained on centralized data in LR, DNN, and SVM (P > 0.05).</p><p><strong>Conclusions: </strong>We demonstrate that the DFL workflow without data sharing should be a more appropriate method in ML tasks in multicentered studies. And the DFL workflow should be further exploited in clinical researches in other departments and it can encourage and facilitate multicentered studies.</p>\",\"PeriodicalId\":23906,\"journal\":{\"name\":\"World neurosurgery\",\"volume\":\" \",\"pages\":\"1036-1046\"},\"PeriodicalIF\":1.9000,\"publicationDate\":\"2025-01-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"World neurosurgery\",\"FirstCategoryId\":\"3\",\"ListUrlMain\":\"https://doi.org/10.1016/j.wneu.2024.10.091\",\"RegionNum\":4,\"RegionCategory\":\"医学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"2024/11/20 0:00:00\",\"PubModel\":\"Epub\",\"JCR\":\"Q3\",\"JCRName\":\"CLINICAL NEUROLOGY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"World neurosurgery","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1016/j.wneu.2024.10.091","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2024/11/20 0:00:00","PubModel":"Epub","JCR":"Q3","JCRName":"CLINICAL NEUROLOGY","Score":null,"Total":0}
引用次数: 0
摘要
背景:分散联合学习(DFL)可作为多中心研究中机器学习(ML)任务的有用框架,在不共享数据的情况下最大限度地利用临床数据。我们的目标是为多中心研究中的机器学习任务提出首个 DFL 工作流程,其功能与使用集中数据的工作流程一样强大:方法:开发的 DFL 工作流程包括四个步骤:注册、局部计算、模型更新和检查。共有 598 名来自 PUMCH 的肢端肥大症患者和 120 名来自 XWH 的患者参与了研究。来自 PUMCH 的队列进一步分为五个中心。九个临床特征被纳入基于四种算法训练的 ML 模型:LR、GBDT、SVM 和 DNN。接受者操作特征曲线的曲线下面积(AUC)用于评估模型的性能:结果:基于 DFL 工作流程训练的模型在 LR 中的表现优于大多数模型(P0.05):我们证明,在多中心研究的 ML 任务中,不共享数据的 DFL 工作流应该是一种更合适的方法。而且,DFL 工作流应在其他部门的临床研究中得到进一步利用,它可以鼓励和促进多中心研究。
Federated Learning for Predicting Postoperative Remission of Patients with Acromegaly: A Multicentered Study.
Background: Decentralized federated learning (DFL) may serve as a useful framework for machine learning (ML) tasks in multicentered studies, maximizing the use of clinical data without data sharing. We aim to propose the first workflow of DFL for ML tasks in multicentered studies, which can be as powerful as those using centralized data.
Methods: A DFL workflow was developed with 4 steps: registration, local computation, model update, and inspection. A total of 598 participants with acromegaly from Peking Union Medical College Hospital, and 120 participants from Xuanwu Hospital were enrolled. The cohort from Peking Union Medical College Hospital was further split into 5 centers. Nine clinical features were incorporated into ML-based models trained based on 4 algorithms: logistic regression (LR), gradient boosted decision tree, support vector machine (SVM), and deep neural network (DNN). The area under the curve of receiver operating characteristic curves was used to evaluate the performance of the models.
Results: Models trained based on DFL workflow performed better than most models in LR (P < 0.05), all models in DNN, SVM, and gradient boosted decision tree (P < 0.05). Models trained on DFL workflow performed as powerful as models trained on centralized data in LR, DNN, and SVM (P > 0.05).
Conclusions: We demonstrate that the DFL workflow without data sharing should be a more appropriate method in ML tasks in multicentered studies. And the DFL workflow should be further exploited in clinical researches in other departments and it can encourage and facilitate multicentered studies.
期刊介绍:
World Neurosurgery has an open access mirror journal World Neurosurgery: X, sharing the same aims and scope, editorial team, submission system and rigorous peer review.
The journal''s mission is to:
-To provide a first-class international forum and a 2-way conduit for dialogue that is relevant to neurosurgeons and providers who care for neurosurgery patients. The categories of the exchanged information include clinical and basic science, as well as global information that provide social, political, educational, economic, cultural or societal insights and knowledge that are of significance and relevance to worldwide neurosurgery patient care.
-To act as a primary intellectual catalyst for the stimulation of creativity, the creation of new knowledge, and the enhancement of quality neurosurgical care worldwide.
-To provide a forum for communication that enriches the lives of all neurosurgeons and their colleagues; and, in so doing, enriches the lives of their patients.
Topics to be addressed in World Neurosurgery include: EDUCATION, ECONOMICS, RESEARCH, POLITICS, HISTORY, CULTURE, CLINICAL SCIENCE, LABORATORY SCIENCE, TECHNOLOGY, OPERATIVE TECHNIQUES, CLINICAL IMAGES, VIDEOS