Poincaré analysis of an overnight arterial oxygen saturation signal applied to the diagnosis of sleep apnea hypopnea syndrome

Combining Heart Rate Variability and Oximetry to Improve Apneic Event Screening in Non-Desaturating Patients

In this paper, we thoroughly analyze the detection of sleep apnea events in the context of Obstructive Sleep Apnea (OSA), which is considered a public health problem because of its high prevalence and serious health implications. We especially evaluate patients who do not always show desaturations during apneic episodes (non-desaturating patients). For this purpose, we use a database (HuGCDN2014-OXI) that includes desaturating and non-desaturating patients, and we use the widely used Physionet Apnea Dataset for a meaningful comparison with prior work. Our system combines features extracted from the Heart-Rate Variability (HRV) and SpO₂, and it explores their potential to characterize desaturating and non-desaturating events. The HRV-based features include spectral, cepstral, and nonlinear information (Detrended Fluctuation Analysis (DFA) and Recurrence Quantification Analysis (RQA)). SpO₂-based features include temporal (variance) and spectral information. The features feed a Linear Discriminant Analysis (LDA) classifier. The goal is to evaluate the effect of using these features either individually or in combination, especially in non-desaturating patients. The main results for the detection of apneic events are: (a) Physionet success rate of 96.19%, sensitivity of 95.74% and specificity of 95.25% (Area Under Curve (AUC): 0.99); (b) HuGCDN2014-OXI of 87.32%, 83.81% and 88.55% (AUC: 0.934), respectively. The best results for the global diagnosis of OSA patients (HuGCDN2014-OXI) are: success rate of 95.74%, sensitivity of 100%, and specificity of 89.47%. We conclude that combining both features is the most accurate option, especially when there are non-desaturating patterns among the recordings under study.

Sleep Apnea Detection Using Multi-Error-Reduction Classification System with Multiple Bio-Signals

INTRODUCTION

Obstructive sleep apnea (OSA) can cause serious health problems such as hypertension or cardiovascular disease. The manual detection of apnea is a time-consuming task, and automatic diagnosis is much more desirable. The contribution of this work is to detect OSA using a multi-error-reduction (MER) classification system with multi-domain features from bio-signals.

METHODS

Time-domain, frequency-domain, and non-linear analysis features are extracted from oxygen saturation (SaO2), ECG, airflow, thoracic, and abdominal signals. To analyse the significance of each feature, we design a two-stage feature selection. Stage 1 is the statistical analysis stage, and Stage 2 is the final feature subset selection stage using machine learning methods. In Stage 1, two statistical analyses (the one-way analysis of variance (ANOVA) and the rank-sum test) provide a list of the significance level of each kind of feature. Then, in Stage 2, the support vector machine (SVM) algorithm is used to select a final feature subset based on the significance list. Next, an MER classification system is constructed, which applies a stacking with a structure that consists of base learners and an artificial neural network (ANN) meta-learner.

RESULTS

The Sleep Heart Health Study (SHHS) database is used to provide bio-signals. A total of 66 features are extracted. In the experiment that involves a duration parameter, 19 features are selected as the final feature subset because they provide a better and more stable performance. The SVM model shows good performance (accuracy = 81.68%, sensitivity = 97.05%, and specificity = 66.54%). It is also found that classifiers have poor performance when they predict normal events in less than 60 s. In the next experiment stage, the time-window segmentation method with a length of 60s is used. After the above two-stage feature selection procedure, 48 features are selected as the final feature subset that give good performance (accuracy = 90.80%, sensitivity = 93.95%, and specificity = 83.82%). To conduct the classification, Gradient Boosting, CatBoost, Light GBM, and XGBoost are used as base learners, and the ANN is used as the meta-learner. The performance of this MER classification system has the accuracy of 94.66%, the sensitivity of 96.37%, and the specificity of 90.83%.

A Hybrid Feature Selection and Extraction Methods for Sleep Apnea Detection Using Bio-Signals

People with sleep apnea (SA) are at increased risk of having stroke and cardiovascular diseases. Polysomnography (PSG) is used to detect SA. This paper conducts feature selection from PSG signals and uses a support vector machine (SVM) to detect SA. To analyze SA, the Physionet Apnea Database was used to obtain various features. Electrocardiography (ECG), oxygen saturation (SaO₂), airflow, abdominal, and thoracic signals were used to provide various frequency-, time-domain and non-linear features (n = 87). To analyse the significance of these features, firstly, two evaluation measures, the rank-sum method and the analysis of variance (ANOVA) were used to evaluate the significance of the features. These features were then classified according to their significance. Finally, different class feature sets were presented as inputs for an SVM classifier to detect the onset of SA. The hill-climbing feature selection algorithm and the k-fold cross-validation method were applied to evaluate each classification performance. Through the experiments, we discovered that the best feature set (including the top-five significant features) obtained the best classification performance. Furthermore, we plotted receiver operating characteristic (ROC) curves to examine the performance of the SVM, and the results showed the SVM with Linear kernel (regularization parameter = 1) outperformed other classifiers (area under curve = 95.23%, sensitivity = 94.29%, specificity = 96.17%). The results confirm that feature subsets based on multiple bio-signals have the potential to identify patients with SA. The use of a smaller subset avoids dimensionality problems and reduces the computational load.

Feature Selection for the Detection of Sleep Apnea using Multi-Bio Signals from Overnight Polysomnography

Patients with sleep apnea (SA) are at increased risk of stroke and cardiovascular disease. Diagnosis of sleep apnea depends on the standard overnight polysomnography (PSG). In this study, the DREAM Apnea Database was used to evaluate the importance of the various features proposed in the literature for the analysis of sleep apnea. Various timeand frequency- domain features that include wavelet and power spectral density were extracted from ECG, EMG, EEG, airflow, SaO2, abdominal and thoracic recordings. Evaluation measures of one-way analysis of variance (ANOVA) and Rank-Sum test were used to test the performance of different features. The selected feature subset indicated that frequency-domain features outperform time-domain ones. This study will help in enhancing the detection accuracy of sleep apnea for the various polysomnography signals.

Noncontact identification of sleep-disturbed breathing from smartphone-recorded sounds validated by polysomnography

PURPOSE

Diagnosis of obstructive sleep apnea by the gold-standard of polysomnography (PSG), or by home sleep testing (HST), requires numerous physical connections to the patient which may restrict use of these tools for early screening. We hypothesized that normal and disturbed breathing may be detected by a consumer smartphone without physical connections to the patient using novel algorithms to analyze ambient sound.

METHODS

We studied 91 patients undergoing clinically indicated PSG. Phase I: In a derivation cohort (n = 32), we placed an unmodified Samsung Galaxy S5 without external microphone near the bed to record ambient sounds. We analyzed 12,352 discrete breath/non-breath sounds (386/patient), from which we developed algorithms to remove noise, and detect breaths as envelopes of spectral peaks. Phase II: In a distinct validation cohort (n = 59), we tested the ability of acoustic algorithms to detect AHI < 15 vs AHI > 15 on PSG.

RESULTS

Smartphone-recorded sound analyses detected the presence, absence, and types of breath sound. Phase I: In the derivation cohort, spectral analysis identified breaths and apneas with a c-statistic of 0.91, and loud obstruction sounds with c-statistic of 0.95 on receiver operating characteristic analyses, relative to adjudicated events. Phase II: In the validation cohort, automated acoustic analysis provided a c-statistic of 0.87 compared to whole-night PSG.

CONCLUSIONS

Ambient sounds recorded from a smartphone during sleep can identify apnea and abnormal breathing verified on PSG. Future studies should determine if this approach may facilitate early screening of SDB to identify at-risk patients for definitive diagnosis and therapy.

CLINICAL TRIALS

NCT03288376; clinicaltrials.org.

Automated detection of sleep apnea in infants using minimally invasive sensors

To address the difficult and necessity of early detection of sleep apnea hypopnea syndrome in infants, we present a study into the effectiveness of pulse oximetry as a minimally invasive means of automated diagnosis of sleep apnea in infants. Overnight polysomnogram data from 328 infants were used to extract time-domain based oximetry features and scored arousal data for each subject. These records were then used to determine apnea events and to train a classifier model based on linear discriminants. Performance of the classifier was evaluated using a leave-one-out cross-validation scheme and an accuracy of 68% was achieved, with a specificity of 68.6% and a sensitivity of 55.9%.

Automated detection of sleep apnea in infants: A multi-modal approach

This study explores the use and applicability of two minimally invasive sensors, electrocardiogram (ECG) and pulse oximetry, in addressing the high costs and difficulty associated with the early detection of sleep apnea hypopnea syndrome in infants. An existing dataset of 396 scored overnight polysomnography recordings were used to train and test a linear discriminants classifier. The dataset contained data from healthy infants, infants diagnosed with sleep apnea, infants with siblings who had died from sudden infant death syndrome (SIDS) and pre-term infants. Features were extracted from the ECG and pulse-oximetry data and used to train the classifier. The performance of the classifier was evaluated using a leave-one-out cross-validation scheme and an accuracy of 66.7% was achieved, with a specificity of 67.0% and a sensitivity of 58.1%. Although the performance of the system is not yet at the level required for clinical use, this work forms an important step in demonstrating the validity and potential for such low-cost and minimally invasive diagnostic systems.

A review of signals used in sleep analysis

This article presents a review of signals used for measuring physiology and activity during sleep and techniques for extracting information from these signals. We examine both clinical needs and biomedical signal processing approaches across a range of sensor types. Issues with recording and analysing the signals are discussed, together with their applicability to various clinical disorders. Both univariate and data fusion (exploiting the diverse characteristics of the primary recorded signals) approaches are discussed, together with a comparison of automated methods for analysing sleep.

Probabilistic neural network approach for the detection of SAHS from overnight pulse oximetry

Diagnosis of sleep apnea hypopnoea syndrome (SAHS) depends on the apnea-hypopnea index determined by the standard in-laboratory overnight polysomnography (PSG). PSG is a costly, labor intensive and, at times, inaccessible approach. Because of the high demand, the need for timely diagnosis and the associated costs, novel methods for SAHS detection are required. In this study, a novel multivariate system is proposed for SAHS detection from the analysis of overnight blood oxygen saturation (SpO2). 115 subjects with SAHS suspicion were studied. A starting set of 17 time domain, stochastic, frequency-domain and nonlinear features were initially computed from SpO2 recordings. Sequential forward feature selection and a probabilistic neural network with leave-one-out cross-validation were applied. Oxygen desaturations below a 4 % threshold within 30 s (ODI430), restorations of 4 % within 10 s (RES4), median value (Sat50), SD1 Poincaré descriptor and the relative power in the 0.013-0.067 Hz frequency band (PSD15/75) formed the optimum features subset. 92.4 % sensitivity and 95.9 % specificity were achieved. Results significantly outperformed the univariate and multivariate approaches reported in literature. The outcome is a simple cost-effective tool that could be used as an alternative or supplementary method in a domiciliary approach to early diagnosis of SAHS.

Automated frequency domain analysis of oxygen saturation as a screening tool for SAHS

Sleep apnea-hypopnea syndrome (SAHS) is significantly underdiagnosed and new screening systems are needed. The analysis of oxygen desaturation has been proposed as a screening method. However, when oxygen saturation (SpO(2)) is used as a standalone single channel device, algorithms working in time domain achieve either a high sensitivity or a high specificity, but not usually both. This limitation arises from the dependence of time-domain analysis on absolute SpO(2) values and the lack of standardized thresholds defined as pathological. The aim of this study is to assess the degree of concordance between SAHS screening using offline frequency domain processing of SpO(2) signals and the apnea-hypopnea index (AHI), and the diagnostic performance of such a new method. SpO(2) signals from 115 subjects were analyzed. Data were divided in a training data set (37) and a test set (78). Power spectral density was calculated and related to the desaturation index scored by physicians. A frequency desaturation index (FDI) was then estimated and its accuracy compared to the classical desaturation index and to the apnea-hypopnea index. The findings point to a high diagnostic agreement: the best sensitivity and specificity values obtained were 83.33% and 80.44%, respectively. Moreover, the proposed method does not rely on absolute SpO(2) values and is highly robust to artifacts.

Wireless portable electrocardiogram and a tri-axis accelerometer implementation and application on sleep activity monitoring

Night-to-night variability of sleep activity requires more home-based portable sleep monitoring instead of clinical polysomnography examination in the laboratory. In this article, a wireless sleep activity monitoring system is described. The system is light and small for the user. Sleep postures, such as supine or left/right side, were observed by a signal from a tri-axis accelerometer. An overnight electrocardiogram was also recorded with a single lead. Using an MSP430 as microcontroller, both physiological signals were transmitted by a Bluetooth chip. A Labview-based interface demonstrated the recorded signal and sleep posture. Three nights of sleep recordings were used to examine night-to-night variability. The proposed system can record overnight heart rate. Results show that sleep posture and posture change can be precisely detected via tri-axis accelerometer information. There is no significant difference within subject data sets, but there are statistically significant differences among subjects, both for heart rate and for sleep posture distribution. The wireless transmission range is also sufficient for home-based users.

Isoflurane-induced changes in righting response and breathing are modulated by RGS proteins

BACKGROUND

Recent evidence suggests that G protein-coupled receptors, especially those linked to G(alpha)(i), contribute to the mechanisms of anesthetic action. Regulator of G protein signaling (RGS) proteins bind to activated G(alpha)(i) and inhibit signal transduction. Genomic knock-in mice with an RGS-insensitive G(alpha)(i2) G184S (G(alpha)(i2) GS) allele exhibit enhanced G(alpha)(i2) signaling and provide a novel approach for investigating the role of G(alpha)(i2) signaling and RGS proteins in general anesthesia.

METHODS

We anesthetized homozygous G(alpha)(i2) GS/GS and wild-type (WT) mice with isoflurane and quantified time (in seconds) to loss and resumption of righting response. During recovery from isoflurane anesthesia, breathing was quantified in a plethysmography chamber for both lines of mice.

RESULTS

G(alpha)(i2) GS/GS mice required significantly less time for loss of righting and significantly more time for resumption of righting than WT mice. During recovery from isoflurane anesthesia, G(alpha)(i2) GS/GS mice exhibited significantly greater respiratory depression. Poincaré analyses show that GS/GS mice have diminished respiratory variability compared with WT mice.

CONCLUSION

Modulation of G(alpha)(i2) signaling by RGS proteins alters loss and resumption of wakefulness and state-dependent changes in breathing.