Sleep disturbance related symptoms are common in patients with long-term oxygen therapy (LTOT). Essentially, there were only few previous reports about the sleep architecture in patients with respiratory disease, such...Sleep disturbance related symptoms are common in patients with long-term oxygen therapy (LTOT). Essentially, there were only few previous reports about the sleep architecture in patients with respiratory disease, such as chronic obstructive pulmonary disease (COPD). This study aims to clarify the objective sleep state and the elements that affect sleep architecture in Chronic Respiratory Failure (CRF) patients with focus on clinical cases of chronic hypercapnia. 13 subjects with chronic respiratory failure were enrolled in the study. All the subjects were pre-evaluated by pulmonary function test and Arterial blood gas analysis (ABG) including exercise testing. Polysomnography (PSG) test was performed in each subject with supplemental oxygen. The estimated base line PaCO2 value that reflects overall PaCO2 including sleep period was calculated using equation of PaCO2[2.4×(HCOˉ3)-22]from obtained ABG value just before PSG test. 6 subjects were classified as hypercapnic group (base line PaCO2 ≥ 45 mmHg) and 7 subjects were non-hypercapnic group (base line PaCO2 < 45 mmHg). Latency persistent sleep of PSG data was significant higher in patients with hypercapnic than non-hypercapnic (p < 0.01). Periodic Limb Movement was seen in 23.6% of the subjects, however there was no contribution for arousals. Other PSG data include mean SpO2 were no significant difference. This study suggests that patients with estimated hypercapnia had more disturbed sleep architecture especially significant loss of sleep latency than non-hypercapnic patient with chronic respiratory failure under LTOT. Nocturnal PaCO2 level or ventilatory function may contribute to sleep disturbance in patients with estimated hypercapnia during LTOT.展开更多
为了更全面地对睡眠脑电进行特征提取,提出一种基于多视图与注意力机制的睡眠脑电分期方法。首先针对原始睡眠脑电信号构造时域和时频域两类视图数据;然后设计融合注意力机制的混合神经网络对多视图数据进行表征学习;接着通过双向长短...为了更全面地对睡眠脑电进行特征提取,提出一种基于多视图与注意力机制的睡眠脑电分期方法。首先针对原始睡眠脑电信号构造时域和时频域两类视图数据;然后设计融合注意力机制的混合神经网络对多视图数据进行表征学习;接着通过双向长短时记忆(bi-directional long short-term memory,BiLSTM)网络进一步学习睡眠阶段之间的转换规则;最后使用Softmax函数进行睡眠分期,并利用类别加权损失函数解决睡眠数据类别不均衡的问题。实验使用Sleep-EDF数据库中前20名受试者的单通道脑电信号并采用20折交叉验证对模型进行性能评估,睡眠分期准确率达到83.7%,宏平均F_(1)值达到79.0%,Cohen′s Kappa系数达到0.78。与现有方法相比,算法性能提升明显,证明了所提方法的有效性。展开更多
文摘Sleep disturbance related symptoms are common in patients with long-term oxygen therapy (LTOT). Essentially, there were only few previous reports about the sleep architecture in patients with respiratory disease, such as chronic obstructive pulmonary disease (COPD). This study aims to clarify the objective sleep state and the elements that affect sleep architecture in Chronic Respiratory Failure (CRF) patients with focus on clinical cases of chronic hypercapnia. 13 subjects with chronic respiratory failure were enrolled in the study. All the subjects were pre-evaluated by pulmonary function test and Arterial blood gas analysis (ABG) including exercise testing. Polysomnography (PSG) test was performed in each subject with supplemental oxygen. The estimated base line PaCO2 value that reflects overall PaCO2 including sleep period was calculated using equation of PaCO2[2.4×(HCOˉ3)-22]from obtained ABG value just before PSG test. 6 subjects were classified as hypercapnic group (base line PaCO2 ≥ 45 mmHg) and 7 subjects were non-hypercapnic group (base line PaCO2 < 45 mmHg). Latency persistent sleep of PSG data was significant higher in patients with hypercapnic than non-hypercapnic (p < 0.01). Periodic Limb Movement was seen in 23.6% of the subjects, however there was no contribution for arousals. Other PSG data include mean SpO2 were no significant difference. This study suggests that patients with estimated hypercapnia had more disturbed sleep architecture especially significant loss of sleep latency than non-hypercapnic patient with chronic respiratory failure under LTOT. Nocturnal PaCO2 level or ventilatory function may contribute to sleep disturbance in patients with estimated hypercapnia during LTOT.
文摘为了更全面地对睡眠脑电进行特征提取,提出一种基于多视图与注意力机制的睡眠脑电分期方法。首先针对原始睡眠脑电信号构造时域和时频域两类视图数据;然后设计融合注意力机制的混合神经网络对多视图数据进行表征学习;接着通过双向长短时记忆(bi-directional long short-term memory,BiLSTM)网络进一步学习睡眠阶段之间的转换规则;最后使用Softmax函数进行睡眠分期,并利用类别加权损失函数解决睡眠数据类别不均衡的问题。实验使用Sleep-EDF数据库中前20名受试者的单通道脑电信号并采用20折交叉验证对模型进行性能评估,睡眠分期准确率达到83.7%,宏平均F_(1)值达到79.0%,Cohen′s Kappa系数达到0.78。与现有方法相比,算法性能提升明显,证明了所提方法的有效性。
