Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2014 Sep 15;9(9):e107581.
doi: 10.1371/journal.pone.0107581. eCollection 2014.

Sleep Apnea-Hypopnea Quantification by Cardiovascular Data Analysis

Affiliations
Free PMC article

Sleep Apnea-Hypopnea Quantification by Cardiovascular Data Analysis

Sabrina Camargo et al. PLoS One. .
Free PMC article

Abstract

Sleep disorders are a major risk factor for cardiovascular diseases. Sleep apnea is the most common sleep disturbance and its detection relies on a polysomnography, i.e., a combination of several medical examinations performed during a monitored sleep night. In order to detect occurrences of sleep apnea without the need of combined recordings, we focus our efforts on extracting a quantifier related to the events of sleep apnea from a cardiovascular time series, namely systolic blood pressure (SBP). Physiologic time series are generally highly nonstationary and entrap the application of conventional tools that require a stationary condition. In our study, data nonstationarities are uncovered by a segmentation procedure which splits the signal into stationary patches, providing local quantities such as mean and variance of the SBP signal in each stationary patch, as well as its duration L. We analysed the data of 26 apneic diagnosed individuals, divided into hypertensive and normotensive groups, and compared the results with those of a control group. From the segmentation procedure, we identified that the average duration <L>, as well as the average variance <σ2>, are correlated to the apnea-hypoapnea index (AHI), previously obtained by polysomnographic exams. Moreover, our results unveil an oscillatory pattern in apneic subjects, whose amplitude S* is also correlated with AHI. All these quantities allow to separate apneic individuals, with an accuracy of at least 79%. Therefore, they provide alternative criteria to detect sleep apnea based on a single time series, the systolic blood pressure.

Conflict of interest statement

Competing Interests: Juergen Kurths and Thomas Penzel are PLOS ONE Editorial Board members. However, this does not alter the authors' adherence to all the PLOS ONE policies on sharing data and materials.

Figures

Figure 1
Figure 1. Systolic blood pressure (SBP) time series (black line) and its local mean values from segmentation (light orange line) for typical hypertensive (upper panel), normotensive (middle panel) and control (lower panel) subjects.
Apnea events, detected via a polysomnography examination, are represented by the light gray vertical lines.
Figure 2
Figure 2. Complementary cumulative distribution of segment lengths for each individual (color) and accumulated data of all subjects in the same group (black solid line), for the hypertensive (HT), normotensive (NT) and control (C) groups.
Drawn for comparison, the thin line in the first two panels (HT and NT) reproduces the accumulated data curve for the control group (C).
Figure 3
Figure 3. Mean length of the segments versus AHI for each subject.
The dashed horizontal line represents the threshold value obtained by a ROC analysis.
Figure 4
Figure 4. Local variance (black lines) provided by the segmentation of SBP and the standard apnea detection represented by the light gray lines, for the same examples of Fig. 1.
Figure 5
Figure 5. Quartiles of the distribution of the average variance and of the average mean, , within each group.
The horizontal lines limit the quartiles, the thicker one indicates the median.
Figure 6
Figure 6. Mean variance of the segments versus AHI for each subject.
The dashed horizontal line represents the ROC threshold.
Figure 7
Figure 7. Systolic blood pressure time series, filtered by subtracting the local mean (black lines), and standard apnea detection (light gray vertical lines).
Figure 8
Figure 8. Autocorrelation of the filtered SBP time series, computed for all the individuals of each group.
Figure 9
Figure 9. Left panels: Patches during non-apnea and apnea epochs (recognizable by the absence/presence of gray vertical lines), for hypertensive and normotensive subjects.
Right panels: the corresponding autocorrelation function for the original and filtered (local mean formula image subtracted) signals.
Figure 10
Figure 10. Power spectrum of the filtered SBP signal, for a typical individual of each group.
Frequency corresponds to inverse interval number.
Figure 11
Figure 11. Normalized maximum of the power spectrum versus AHI, for each individual.
The dashed horizontal line represents the ROC threshold.
Figure 12
Figure 12. ROC curves for , and , obtained to identify patients with apnea.
Threshold values (21.4, 112 and 39, respectively) shown in previous figures were obtained from the optimal classification corresponding to the lowest distance to the upper left corner.
Figure 13
Figure 13. Normalized maximum of the spectral density vs mean segment length , for each individual. Dashed lines represent threshold values.

Similar articles

See all similar articles

Cited by 2 articles

References

    1. Penzel T, Wessel N, Riedl M, Kantelhardt JW, Rostig S, et al. (2007) Cardiovascular and respiratory dynamics during normal and pathological sleep. Chaos 17: 015116. - PubMed
    1. Stickgold R (2005) Sleep-dependent memory consolidation. Nature 437: 1272. - PubMed
    1. Smith C (1995) Sleep states and memory processes. Behav Brain Res 69: 137. - PubMed
    1. Penzel T, McNames J, de Chazal P, Raymond B, Murray A, et al. (2002) Systematic comparison of different algorithms for apnoea detection based on electrocardiogram recordings. MEDICAL & BIOLOGICAL ENGINEERING & COMPUTING 40: 402–407. - PubMed
    1. Helland VCF, Gapelyuk A, Suhrbier A, Riedl M, Penzel T, et al. (2010) Investigation of an automatic sleep stage classification by means of multiscorer hypnogram. Methods Inf Med 49: 467. - PubMed

Publication types

Grant support

Funding provided by CNPq (Brazilian agency) http://www.cnpq.br/ and Deutsche Forschungsgemeinschaft, grant numbers DFG RI2016/2-1, WE 2834/5-1, KU837/29-2, and KU837/35-1. http://www.dfg.de/. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Feedback