[论文解读] Sleep Staging by Modeling Sleep Stage Transitions using Deep CRF.

Karan Aggarwal, Swaraj Khadanga+3 more|arXiv (Cornell University)|Jul 23, 2018

Obstructive Sleep Apnea Research参考文献 41被引用 2

一句话总结

本文提出了一种端到端的深度学习框架，仅使用CPAP通气流量信号对睡眠阶段转换进行建模，结合卷积-循环神经网络进行特征提取，并引入深层条件随机场（CRF）层以捕捉时序动态。该方法相比先前的方法将睡眠分期准确率提高了10%，并实现了对CPAP治疗过程中睡眠效率和阶段指标的可靠追踪。

ABSTRACT

Sleep plays a vital role in human health, both mental and physical. Sleep disorders like sleep apnea are increasing in prevalence, with the rapid increase in factors like obesity. Sleep apnea is most commonly treated with Continuous Positive Air Pressure (CPAP) therapy, which maintains the appropriate pressure to ensure continuous airflow. It is widely accepted that in addition to preventing air passage collapse, increase in deep and REM sleep stages would be good metrics for how well the CPAP therapy is working in improving sleep health. Presently, however, there is no mechanism to easily detect a patient's sleep stages from CPAP flow data alone. We propose, for the first time, an automated sleep staging model based only on the flow signal. Recently deep neural networks have shown high accuracy on sleep staging by eliminating handcrafted features. However, these methods focus exclusively on extracting informative features from the input signal, without paying much attention to the dynamics of sleep stages in the output sequence. We propose an end-to-end framework that uses a deep convolution-recurrent neural networks to extract high-level features from raw flow signal and then uses a structured output layer based on a conditional random field to model the temporal transition structure of the sleep stages. We improve upon the previous methods by 10% using our model, that can be augmented to the previous sleep staging deep learning methods. We also show that our method can be used to accurately track sleep metrics like sleep efficiency calculated from sleep stages that can be deployed for monitoring the response of CPAP therapy on sleep apnea patients. Apart from the technical contributions, we expect this study to motivate new research questions in sleep science, especially towards the understanding of sleep architecture trajectory among patients under CPAP therapy.

研究动机与目标

• 开发一种仅依赖CPAP流量信号的自动化睡眠分期系统，无需多导睡眠图或额外传感器。
• 解决现有深度学习模型在睡眠分期中忽略睡眠阶段转换时序结构的局限性。
• 通过结构化预测建模睡眠阶段的序列动态，提升睡眠分期的准确性。
• 通过推导出的睡眠效率和阶段持续时间等临床指标，实现对CPAP治疗效果的临床监测。

提出的方法

• 采用深度卷积-循环神经网络（CRNN）从原始CPAP流量信号中直接提取分层的高层次表征。
• 将条件随机场（CRF）层作为结构化输出层，以建模睡眠阶段之间的时序依赖关系和转换概率。
• 使用联合损失函数端到端训练整个框架，结合特征学习与序列级结构化预测。
• 利用预测的睡眠阶段序列计算临床相关的指标，如睡眠效率、深度睡眠和REM睡眠持续时间。
• 通过添加CRF层增强现有深度学习睡眠分期模型，提升性能，且无需重新训练特征提取器。
• 利用CRF的时序建模能力，强制实现生理上合理的睡眠阶段转换，如从N1到N2再到SWS的进展。

实验结果

研究问题

• RQ1 是否可以仅使用CPAP流量信号而无需其他生理信号，实现准确的睡眠分期？
• RQ2 通过结构化CRF层建模睡眠阶段转换，与标准深度学习方法相比，如何提升分期准确率？
• RQ3 该模型在仅依赖流量数据的情况下，能在多大程度上追踪睡眠效率和REM睡眠持续时间等临床相关睡眠指标？
• RQ4 该模型能否与现有深度学习睡眠分期框架集成以提升性能？
• RQ5 该模型为CPAP治疗期间睡眠结构变化提供了哪些新见解？

主要发现

• 在CPAP流量数据上评估时，所提模型相比先前的深度学习方法，睡眠分期准确率提高了10%。
• 集成深层CRF层显著提升了性能，通过建模睡眠阶段转换的时序动态。
• 该模型可仅从流量信号中准确估算睡眠效率和阶段持续时间，支持对CPAP治疗的临床监测。
• 该框架可无缝集成到现有深度学习睡眠分期模型中，提升其性能，且无需重新训练特征提取器。
• 该方法为研究接受CPAP治疗患者睡眠结构轨迹提供了基础，为睡眠科学开辟了新的研究方向。