Privacy-first health research with federated learning.
Privacy-first health research with federated learning.
复制标题
DOI：
10.1038/s41746-021-00489-2
复制
发表时间：
2021-09-07
影响因子：
15.2
通讯作者：
Hernandez J
中科院分区：
文献类型：
--
作者：
Sadilek A;Liu L;Nguyen D;Kamruzzaman M;Serghiou S;Rader B;Ingerman A;Mellem S;Kairouz P;Nsoesie EO;MacFarlane J;Vullikanti A;Marathe M;Eastham P;Brownstein JS;Arcas BAY;Howell MD;Hernandez J
隐私保护在开展健康研究中至关重要。然而，研究通常依赖于存储在集中存储库中的数据，在那里进行分析时可以完全访问敏感的底层内容。联邦学习的最新进展使构建以分布式方式训练的复杂机器学习模型成为可能。这些技术促进了研究端点的计算，这样私有数据就不会离开给定的设备或医疗保健系统。我们通过一组不同的单点和多点健康研究表明，联邦模型可以达到类似的准确性、精确度和泛化性，并产生与标准集中式统计模型相同的解释，同时实现相当强的隐私保护，并且不会显著提高计算成本。这项工作是第一个应用现代和通用的联邦学习方法，明确地将差异隐私纳入临床和流行病学研究——跨越联邦单元、模型架构、学习任务的复杂性和疾病的范围。因此，它使健康研究参与者能够继续控制他们的数据，并仍然为过去相互矛盾的科学方面的进步做出贡献。
Privacy protection is paramount in conducting health research. However, studies often rely on data stored in a centralized repository, where analysis is done with full access to the sensitive underlying content. Recent advances in federated learning enable building complex machine-learned models that are trained in a distributed fashion. These techniques facilitate the calculation of research study endpoints such that private data never leaves a given device or healthcare system. We show—on a diverse set of single and multi-site health studies—that federated models can achieve similar accuracy, precision, and generalizability, and lead to the same interpretation as standard centralized statistical models while achieving considerably stronger privacy protections and without significantly raising computational costs. This work is the first to apply modern and general federated learning methods that explicitly incorporate differential privacy to clinical and epidemiological research—across a spectrum of units of federation, model architectures, complexity of learning tasks and diseases. As a result, it enables health research participants to remain in control of their data and still contribute to advancing science—aspects that used to be at odds with each other.
登录
查看更多内容
影响因子：
3.2
作者：
Lee J;Sun J;Wang F;Wang S;Jun CH;Jiang X
通讯作者：
Jiang X
影响因子：
3.7
作者：
Ohene, Sally-Ann;Bakker, Mirjam I.;Klatser, Paul
通讯作者：
Klatser, Paul
影响因子：
9.8
作者：
Johnson AE;Pollard TJ;Shen L;Lehman LW;Feng M;Ghassemi M;Moody B;Szolovits P;Celi LA;Mark RG
通讯作者：
Mark RG
影响因子：
3.5
作者：
Chicco, Davide;Jurman, Giuseppe
通讯作者：
Jurman, Giuseppe
影响因子：
22.7
作者：
Rugge, Massimo;Zorzi, Manuel;Guzzinati, Stefano
通讯作者：
Guzzinati, Stefano