SCH: INT: Collaborative Research: Privacy-Preserving Federated Transfer Learning for Early Acute Kidney Injury Risk Prediction

SCH：INT：合作研究：用于早期急性肾损伤风险预测的隐私保护联合迁移学习

• 批准号: 2014554
• 负责人: Mei Liu
• 金额: $ 61.9万
• 依托单位: University of Kansas Medical Center Research Institute Inc
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2020
• 资助国家: 美国
• 起止时间: 2020-10-01 至 2024-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2014554&HistoricalAwards=false
• 关键词: SCH; INT; Collaborative; Research; Privacy

Federated learning enables hospitals to collaboratively learn a shared global model while ensuring patient privacy; however, there is a big statistical challenge for our application owing to EHR heterogeneities, i.e. difference in patient characteristics and clinical observations made or feature space. Thus, real-world EHR data from different hospitals are never independently and identically distributed (IID). The proposed research is to overcome this statistical challenge while improving security for federated learning byleveraging a large integrated EHR dataset with medical records for more than 21 million patients from 12 healthcare systems spanning across 9 US states. A novel privacy-preserving federated transfer learning framework is proposed for building a robust and accurate AKI prediction model that require learning on real-world EHR data from siloed healthcare systems. This project will (1) develop novel transfer learning solutions to address three distinct non-IID EHR data analytic scenarios, (2) develop a novel federated learning framework with a dynamic weighting aggregation mechanism to build a robust and accurate Acute kidney injury (AKI) prediction model; and (3) develop a comprehensive privacy-preserving federated transfer learning framework with novel privacy-preserving solutions to address the unique privacy challenges in the proposed transfer learning applications.The project proposes new transfer learning solutions to combat the non-IID challenge in federated learning and new security building blocks tailored for homogeneous and heterogeneous transfer learning tasks. Together the project will develop a privacy-preserving federated transfer learning framework to provide a first practical solution for non-IID clinical data scenarios. Our research methods and findings will provide promising new directions to machine learning for healthcare and will contribute to both academic research and potential commercialized products. More importantly, the interpretable nature of the base gradient boosting machine model in the proposed federated transfer learning framework will provide better understanding of the predictors from which clinicians can use to design prevention and management strategies for high-risk patients.This project is jointly funded by Smart and Connected Health and the Established Program to Stimulate Competitive Research (EPSCoR).This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

联邦学习使医院能够在确保患者隐私的同时协作学习共享的全局模型;然而，由于EHR的异质性，即患者特征和临床观察或特征空间的差异，我们的应用面临着巨大的统计挑战。因此，来自不同医院的真实世界EHR数据从来都不是独立同分布的（IID）。拟议的研究旨在克服这一统计挑战，同时通过利用大型集成EHR数据集来提高联邦学习的安全性，该数据集包含来自美国9个州12个医疗保健系统的2100多万患者的医疗记录。提出了一种新的隐私保护联邦迁移学习框架，用于构建一个强大而准确的阿基预测模型，该模型需要学习来自孤立医疗系统的真实世界EHR数据。该项目将（1）开发新的迁移学习解决方案，以解决三种不同的非IID EHR数据分析场景，（2）开发具有动态加权聚合机制的新的联邦学习框架，以建立强大且准确的急性肾损伤（阿基）预测模型;以及（3）开发一个全面的隐私保护联邦迁移学习框架，该项目提出了新的迁移学习解决方案，以应对联邦学习中的非IID挑战，并为同质和异构迁移学习任务量身定制了新的安全构建模块。该项目将共同开发一个保护隐私的联合迁移学习框架，为非IID临床数据场景提供第一个实用的解决方案。我们的研究方法和发现将为医疗保健机器学习提供有前途的新方向，并将有助于学术研究和潜在的商业化产品。更重要的是，在拟议的联邦迁移学习框架中，基础梯度提升机器模型的可解释性将使临床医生更好地了解预测因素，从而为高风险患者设计预防和管理策略。该项目由智能与互联健康和刺激竞争研究的既定计划（EPSCoR）联合资助该奖项反映了NSF的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

项目成果

Interpretable Sub-phenotype Identification in Acute Kidney Injury

急性肾损伤中可解释的亚表型识别

• 发表时间: 2022
• 期刊: American Medical Informatics Association (AMIA
• 作者: Chan, Ho Yin;Liu, Mei
• 通讯作者: Liu, Mei

Development and Validation of a Personalized Model With Transfer Learning for Acute Kidney Injury Risk Estimation Using Electronic Health Records.

• DOI: 10.1001/jamanetworkopen.2022.19776
• 发表时间: 2022-07-01
• 期刊: JAMA network open
• 影响因子: 13.8

Digital health and acute kidney injury: consensus report of the 27th Acute Disease Quality Initiative workgroup

• DOI: 10.1038/s41581-023-00744-7
• 发表时间: 2023-08-14
• 期刊: NATURE REVIEWS NEPHROLOGY
• 影响因子: 41.5
• 作者: Kashani，Kianoush B.;Awdishu，Linda;Mehta，Ravindra L.
• 通讯作者: Mehta，Ravindra L.

A hybrid adaptive approach for instance transfer learning with dynamic and imbalanced data.

• DOI: 10.1002/int.23055
• 发表时间: 2022-12
• 期刊: INTERNATIONAL JOURNAL OF INTELLIGENT SYSTEMS
• 影响因子: 7
• 作者: Zhang, Xiangzhou;Liu, Kang;Yuan, Borong;Wang, Hongnian;Chen, Shaoyong;Xue, Yunfei;Chen, Weiqi;Liu, Mei;Hu, Yong
• 通讯作者: Hu, Yong

Characterizing the temporal changes in association between modifiable risk factors and acute kidney injury with multi-view analysis

• DOI: 10.1016/j.ijmedinf.2022.104785
• 发表时间: 2022-04
• 期刊: International journal of medical informatics
• 影响因子: 4.9
• 作者: Kang Liu;Borong Yuan;Xiangzhou Zhang;Weiqi Chen;L. Patel;Yong Hu;Mei Liu