虚假阳性耐受模型的不当行为减缓在分布式联合学习电子健康记录数据在临床机构的电子健康记录数据
Maxim Edelson1, Anh Pham2,3, Tsung-Ting Kuo4,5,6
1Department of Computer Science and Engineering, University of California San Diego, La Jolla, CA, USA.
Scientific reports
|July 2, 2025
概括
我们开发了一种假阳性容忍方法,用于检测联邦学习中的模型不当行为,改进了健康预测模型. 这种方法提高了安全性,没有过度的错误报警,保护协作医疗保健数据.
科学领域:
- 计算机科学 计算机科学
- 机器学习 机器学习
- 网络安全 网络安全
背景情况:
- 协作机器学习 (ML) 对于跨机构的健康预测建模至关重要.
- 联合学习 (FL) 网络面临安全风险,包括被改的模型注入 (模型不当行为).
- 现有的检测不当行为的方法往往会产生过度的错误警报,阻碍合作.
研究的目的:
- 提出一个假阳性宽容的方法,以维护联邦学习中的模型完整性.
- 为了减轻分布式ML网络中对抗性不当行为的影响.
- 在协作医疗预测建模中增强安全性.
主要方法:
- 开发了一个假阳性宽容的方法来检测FL的模型不当行为.
- 只有在超过一个门后,才对隔离站点实施了不良行为预算系统.
- 在一个分散的区块链网络上使用三个数据集对系统进行了评估.
主要成果:
- 与非耐受性方法相比,拟议的方法在接收器运行特征曲线 (0.058-0.121) 下面面积显著增加.
- 实现了微不足道的开销,不到12毫秒.
- 有效地保护模型完整性,同时减轻对手的不当行为.
结论:
- 假阳性耐受性方法为保护分布式FL流程提供了一种高效,强大的解决方案.
- 该方法继承了区块链的特点,如透明度和去中心化,以增强对医疗保健的信任.
- 未来的工作包括将该方法应用于更复杂的ML算法和大规模场景.
相关概念视频
Ethical Standards I
985
The American Nurses Association (ANA) created and implemented the first nationally accepted Code of Ethics for Nurses with Interpretive Statements. The Code of Ethics is a living document regularly updated by the ANA and establishes an ethical standard that is non-negotiable for nurses in all roles and settings.
The Code of Ethics provisions outline the nurse's duty to the patient, the healthcare team, the profession, and society. The Code's fundamental principles include advocacy,...
The Code of Ethics provisions outline the nurse's duty to the patient, the healthcare team, the profession, and society. The Code's fundamental principles include advocacy,...
985
Methods of Documentation VII: EMR
921
Electronic Medical Records (EMRs) primarily center around electronically documenting patients' health information within a single healthcare organization or practice. They contain essential clinical data related to a patient's medical history, diagnoses, medications, treatment plans, lab results, and other pertinent information relevant to the specific encounter or episode of care. EMRs are designed to streamline documentation and workflow processes within individual healthcare...
921
Model Approaches for Pharmacokinetic Data: Distributed Parameter Models
131
Pharmacokinetic models are mathematical constructs that represent and predict the time course of drug concentrations in the body, providing meaningful pharmacokinetic parameters. These models are categorized into compartment, physiological, and distributed parameter models.
The distributed parameter models are specifically designed to account for variations and differences in some drug classes. This model is particularly useful for assessing regional concentrations of anticancer or...
The distributed parameter models are specifically designed to account for variations and differences in some drug classes. This model is particularly useful for assessing regional concentrations of anticancer or...
131


