AI 中文总结
研究消费级可穿戴设备通过多种信号推断睡眠阶段的性能，引入四层受控分解框架评估，对比不同信号源的效果，发现非脑电图睡眠分期较粗略，基于置信度弃权可校准操作模式，量化了可穿戴信号与现实传感约束的惩罚。
AI 中文摘要
消费级可穿戴设备越来越多地通过心率、加速度测量和光电容积脉搏波描记术等信号来推断睡眠阶段。然而，现有研究通常在固定信号设置下报告端到端性能，难以确定观察到的性能是来自真正的生理解码、时间先验还是特定数据集的混杂因素。为解决这一局限性，我们引入了一个用于非脑电图睡眠分期的四层受控分解框架，涵盖信号源、生理表征、时间先验和决策层。该框架在跨越苹果手表睡眠加速度计（N = 31）、睡眠心脏健康研究（N = 195，实验室心电图、呼吸和SpO₂信号）以及作为脑电图+眼电图参考的Sleep-EDF-20的信号质量阶梯上进行评估，始终使用相同的紧凑Mamba2模型。实验室心肺信号达到κ = 0.492，而脑电图+眼电图达到κ = 0.796，留下Δκ = +0.304的残余差距，这反映了缺失的皮质信息而非仅时间建模。消费级心率/加速度仅达到κ = 0.255，量化了衍生可穿戴信号和现实世界传感约束的额外惩罚。基于置信度的弃权提供了一种校准操作模式：去除20%最低置信度的时段可将κ从0.452提高到0.512，而标签打乱的对照组则降至κ = -0.003。这些结果支持非脑电图睡眠分期作为粗略的、有置信度意识的睡眠结构监测，而非等同于脑电图的五类临床分期。
英文摘要
Consumer wearables increasingly infer sleep stages from signals including heart rate, accelerometry, and photoplethysmography. However, existing studies often report end-to-end performance under a fixed signal setting, making it difficult to determine whether the observed performance comes from genuine physiological decoding, temporal priors, or dataset-specific confounds. To address this limitation, we introduce a four-layer controlled decomposition framework for non-EEG sleep staging, covering signal source, physiological representation, temporal prior, and decision layers. The framework is evaluated across a signal-quality ladder spanning Apple Watch Sleep-Accel ($N=31$), the Sleep Heart Health Study ($N=195$, laboratory ECG, respiratory, and SpO$_2$ signals), and Sleep-EDF-20 as an EEG+EOG reference, using the same compact Mamba2 model throughout. Laboratory cardiorespiratory signals reach $κ=0.492$, while EEG+EOG reaches $κ=0.796$, leaving a residual gap of $Δκ=+0.304$ that reflects missing cortical information rather than temporal modeling alone. Consumer HR/ACC reaches only $κ=0.255$, quantifying the additional penalty of derived wearable signals and real-world sensing constraints. Confidence-based abstention provides a calibrated operating mode: removing the 20% lowest-confidence epochs increases $κ$ from $0.452$ to $0.512$, while a label-shuffled control collapses to $κ=-0.003$. These results support non-EEG sleep staging as coarse, confidence-aware sleep-structure monitoring rather than EEG-equivalent five-class clinical staging.
Comments9 pages, 5 figures, 6 tables
