对非独立且相同分布的睡眠数据的联合学习进行分析
Adriana Anido Alonso1, Diego Alvarez-Estevez1
1Universidade da Coruña, Campus Elviña s/n, A Coruña, A Coruña, 15006, Spain.
Physiological measurement
|February 25, 2026
概括
联合学习 (FL) 有效地训练睡眠分期模型在不同的非IID数据上,而不影响隐私. 一种新的子样本策略增强了概括性,并减少了计算负载.
科学领域:
- 人工智能的人工智能
- 机器学习 机器学习
- 睡眠科学 睡眠科学
背景情况:
- 集中式睡眠分期模型面临着敏感患者数据的隐私挑战.
- 联合学习 (FL) 提供了一种分散的方法,可以在多个站点之间培训模型,而无需共享数据.
- 不同质的,非独立的,相同分布的 (非IID) 数据在现实应用中对FL构成重大障碍.
研究的目的:
- 评估联合学习 (FL) 算法 (FedSGD,FedAvg,FedProx) 在异质,非IID睡眠数据上的有效性.
- 评估当地培训时代和聚合策略对模型性能的影响.
- 引入和验证一个通用的子样本策略,以减轻数据不平衡,并改善FL中的通用化.
主要方法:
- 采用双层评估框架,评估地方培训时期 (1, 30) 和聚合方案 (加权,未加权).
- 引入了通用子样本策略,以解决跨客户的数据异质性和体积不平衡问题.
- 利用了六个独立的睡眠数据库,并进行了离开一个数据库的交叉验证,以实现强大的外部概括.
主要成果:
- 增加本地培训时间对所有FL计划的业绩产生了负面影响,加剧了客户的偏移.
- 不加权的聚合效果优于加权的聚合效果,这表明客户贡献不成比例的偏差.
- 拟议的分样采集策略显示出最有效的效果,产生一致的概括和减少计算开销.
结论:
- 联合学习 (FL) 可以实现与自动睡眠分期的集中方法相匹配的性能,同时保持数据隐私.
- 分散的培训策略有效地克服了多中心睡眠研究中的数据孤岛.
- 开发的分样采样策略对于在临床环境中强大有效地实施FL至关重要.
相关概念视频
01:11Understanding Sleep
1.7K
Sleep, an essential biological state, involves significant reductions in physical activity, sensory awareness, and interaction with the environment. This complex physiological process is primarily regulated by specific brain regions, notably the hypothalamus and pons, which govern the sleep-wake cycle or circadian rhythm.
The circadian rhythm, a nearly 24-hour cycle, is deeply influenced by environmental light cues. Light exposure directly affects the hypothalamus, which in turn regulates...
The circadian rhythm, a nearly 24-hour cycle, is deeply influenced by environmental light cues. Light exposure directly affects the hypothalamus, which in turn regulates...
1.7K
01:13Insufficient Sleep and Sleep Deprivation
1.0K
Insufficient sleep refers to not getting the recommended amount of sleep for optimal functioning, even if it's just slightly less than needed. Sleep insufficiency may occur due to lifestyle choices, such as staying up late for social events or work, resulting in routinely getting less sleep than required. For example, consistently sleeping 6 hours when the body needs 7-9 hours can lead to cumulative effects on health and well-being.
Sleep deprivation is a more severe form of sleep loss...
Sleep deprivation is a more severe form of sleep loss...
1.0K
01:21Friedman Two-way Analysis of Variance by Ranks
522
Friedman's Two-Way Analysis of Variance by Ranks is a nonparametric test designed to identify differences across multiple test attempts when traditional assumptions of normality and equal variances do not apply. Unlike conventional ANOVA, which requires normally distributed data with equal variances, Friedman's test is ideal for ordinal or non-normally distributed data, making it particularly useful for analyzing dependent samples, such as matched subjects over time or repeated measures...
522
01:24Substance Use Disorders Affecting Sleep
484
Substance use disorders involve a pattern of using drugs more extensively than intended and continuing use despite harmful consequences. This includes legal substances like alcohol and nicotine, as well as illegal drugs. These disorders often involve both physical and psychological dependence, reflecting compulsive use of substances that significantly alter thoughts, feelings, and behaviors, contributing to a major public health issue.
Understanding the concepts of physical dependence,...
Understanding the concepts of physical dependence,...
484