Z. Shinar

Obstructive sleep apnea detection based on electrocardiogram analysis

Obstructive Sleep Apnea is a frequent disorder with detrimental health, performance and safety effects. The diagnosis of the disorder is cumbersome and expensive. New methods for screening and diagnosis are needed. The method the authors describe in this work is based on detection of respiratory disturbance during sleep from continuously monitored ECG during the night. The detection is based on QRS complex changes caused by apneas and spectral abnormalities in the Heart Rate Variability which are related to recurrent respiratory events. An automatic step-by-step approach, based on these two already described phenomena, was developed, validated and then applied on data supplied by Physionet for the CinC Challenge 2000. The methods yield results in excellent agreement with the manually scored standard sleep studies, and achieved 14788/17262 correct classifications on minute-by-minute bases.

Electrocardiogram derived respiration during sleep

The aim of this study was to quantify the ECG Derived Respiration (EDR) in order to extend the capabilities of ECG-based sleep analysis. We examined our results in normal subjects and in patients with Obstructive Sleep Apnea Syndrome (OSAS) or Central Sleep Apnea. Lead 2 ECG and three measures of respiration (thorax and abdominal effort, and oronasal flow signal) were recorded during sleep studies of 12 normal and 12 OSAS patients. Three parameters, the R-wave amplitude (RWA), R-wave duration (RWD), and QRS area, were extracted from the ECG signal, resulting in time series that displayed a behavior similar to that of the respiration signals. EDR frequency was correlated with directly measured respiratory frequency, and averaged over all subjects. The peak-to-peak value of the EDR signals during the apnea event was compared to the average peak-to-peak of the sleep stage, containing the apnea

Automatic detection of slow-wave-sleep using heart rate variability

In this study, we used heart rate variability parameters to first characterize and then automatically detect slow-wave sleep (SWS). First, a wavelet transform was used to decompose equally sampled R-R interval series into their time-dependent spectral components: very low frequency (VLF) 0.005-0-04Hz, low frequency (LF) 0.04-0.15 Hz, and high frequency (HF) 0.15-0.45Hz. Then, the known decrease in LF power during SWS was confirmed and a linear relation between the average LF/HF balance throughout the night and the balance during SWS was found. Also, similar behaviour was found with the VLF power and the VLF/HF ratio. Finally, a decision algorithm with two criteria was defined using a training set of ECG recordings and applied to a test set. The results amounted to an 80% correct identification of SWS. The limitations of the study, as well as inherent differences between SWS definitions based on EEG and ECG, are discussed.

Validation of ECG-derived sleep architecture and ventilation in sleep apnea and chronic fatigue syndrome

PurposeNewly developed algorithms putatively derive measures of sleep, wakefulness, and respiratory disturbance index (RDI) through detailed analysis of heart rate variability (HRV). Here, we establish levels of agreement for one such algorithm through comparative analysis of HRV-derived values of sleep–wake architecture and RDI with those calculated from manually scored polysomnographic (PSG) recordings.MethodsArchived PSG data collected from 234 subjects who participated in a 3-day, 2-night study characterizing polysomnographic traits of chronic fatigue syndrome were scored manually. The electrocardiogram and pulse oximetry channels were scored separately with a novel scoring algorithm to derive values for wakefulness, sleep architecture, and RDI.ResultsFour hundred fifty-four whole-night PSG recordings were acquired, of which, 410 were technically acceptable. Comparative analyses demonstrated no difference for total minutes of sleep, wake, NREM, REM, nor sleep efficiency generated through manual scoring with those derived through HRV analyses. When NREM sleep was further partitioned into slow-wave sleep (stages 3–4) and light sleep (stages 1–2), values calculated through manual scoring differed significantly from those derived through HRV analyses. Levels of agreement between RDIs derived through the two methods revealed an R = 0.89. The Bland–Altman approach for determining levels of agreement between RDIs generated through manual scoring with those derived through HRV analysis revealed a mean difference of −0.7 ± 8.8 (mean ± two standard deviations).ConclusionWe found no difference between values of wakefulness, sleep, NREM, REM sleep, and RDI calculated from manually scored PSG recordings with those derived through analyses of HRV.

Detection of different recumbent body positions from the electrocardiogram

Changes in body position alter the relative angle between ECG electrodes and the mean electric axis of the heart. These changes influence the time interval during which the projection of the electric dipole, on any ECG lead, is positive (R-wave). In this study, measurements of R-wave duration (RWD) were used to identify changes in body position, and two of its uncorrelated features were used to classify each heartbeat into four basic groups relating to four body positions (supine, prone, left-side, right-side). Data were acquired from healthy volunteers during controlled condition experiments that included well-defined sequences of body positions and simultaneous recordings of ECG leads I, II and III. Results showed over 90% correct identifications ofbody position changes when using any of the three leads. Lead II had the best performance for theclassification of body position and correctly classified 80% of heartbeats. Classification did not improve for a combination of two leads. The technique can be used to reveal additional important clinical information and can be easily implemented, in a variety of applications where ECG is recorded, such as sleep studies, Holter recordings and ischaemia detection.

Changes in autonomic nervous system activity and in electro-cortical activity during sleep onset

This study is aimed at investigating the behavior of the autonomic nervous system (ANS) at sleep onset (SO) and to correlate it with higher brain activity captured as surface electroencephalogram (EEG). The study was performed on data obtained from 16 healthy subjects who underwent standard polysomnography (PSG). The instantaneous heart rate variability (HRV) provided a measure of ANS activity, and EEG was quantified by time-frequency analysis measuring power in alpha and delta frequency bands. The results revealed a power decrease in the VLF component of the inter-beat (RR) interval to a minimal level around SO, similar changes, though to a lesser degree in the LF component, and almost no change in the HF component. These changes correlated with a drop in alpha and a surge in delta EEG power. The sympathovagal balance, as reflected by LF/HF ratio, pointed towards an increase in the parasympathetic activity after SO.

R wave duration as a measure of body position changes during sleep

Any change in body position alters the relative angle between the ECG electrodes and the electric axis of the heart. Thus it influences the time interval during which the projection of the electric dipole, on any ECG lead is positive. We used the R wave duration (RWD) as representing this positive projection and as indicator of body position changes. Two experimental setups were used: (1) Simultaneous ECG recordings from leads I, II, and III during a sequence of well-defined body positions, (controlled conditions). (2) Lead II and video recordings from healthy subjects, who underwent whole night sleep studies. Using data of any of the leads, we could identify over 90% of the changes in body position, in both experimental setups. RWD calculated from lead II was the most reproducible upon reassuming the same body position. Measuring RWD from ECG may help to reduce the number of sensors required in full sleep studies, and to obtain better interpretation of non-attended ones.

Cluster headache patients have normal circadian and sleep time autonomic nervous system function

Cluster headache (CH) is a rare form of primary headache of neurovascular origin causing severe pain attacks associated with autonomic changes. Attacks are more likely to occur during sleep. Time-frequency decomposition (TFD) of instantaneous heart rate variability (HRV) is widely accepted as a non-invasive tool of investigation of autonomic nervous function and was applied in the present study. Our goal was to estimate the autonomic features of CH patients and their connection to sleep. The study included 20 subjects belonging to 3 groups: (a) CH (active headache attacks, N=7); (b) Normal control (C, N=6); (c) patients with CH during a quiet period (QP, N=7). The study revealed similar circadian behaviour of all HRV variables and of the HR in all groups indicating normal changes in central autonomic function between daytime and sleep in CH. Increased overall VLF power in CH compared to normal subjects suggests increased vasomotor activity during active headache periods only

Impaired autonomic balance during sleep in obstructive sleep apnea: origin or result

The ECG may be used as a means to uncover information on the function of organs and systems apart from the heart itself. We studied the autonomic function in adult patients with obstructive sleep apnea syndrome (OSAS) and expected their autonomic balance to be shifted towards sympathetic predominance. The study included 12 OSAS patients and 12 subjects with no respiratory disturbance during sleep, as diagnosed by attended whole-night polysomnography (PSG) and scored according to standard criteria. The time-frequency decomposition of beat-to-beat heart rate variability (HRV), detected from the ECG, served to evaluate autonomic function. The results showed an overall increased sympathetic activity during sleep and sympathetic predominance during slow-wave sleep (SWS) in patients, as compared to control subjects. The degree of sympathetic predominance correlated well with the severity of sleep apnea.