Mechanical deconditioning of the heart due to long-term bed rest as observed on seismocardiogram morphology.

During head-down tilt bed rest (HDT) the cardiovascular system is subject to headward fluid shifts. The fluid shift phenomenon is analogous to weightlessness experienced during spaceflight microgravity. The purpose of this study was to investigate the effect of prolonged 60-day bed rest on the mechanical performance of the heart using the morphology of seismocardiography (SCG). Three-lead electrocardiogram (ECG), SCG and blood pressure recordings were collected simultaneously from 20 males in a 60-day HDT study (MEDES, Toulouse, France). The study was divided into two campaigns of ten participants. The first commenced in January, and the second in September. Signals were recorded in the supine position during the baseline data collection (BDC) before bed rest, during 6° HDT bed rest and during recovery (R), post-bed rest. Using SCG and blood pressure at the finger, the following were determined: Pulse Transit Time (PTT); and left-ventricular ejection time (LVET). SCG morphology was analyzed using functional data analysis (FDA). The coefficients of the model were estimated over 20 cycles of SCG recordings of BDC12 and HDT52. SCG fiducial morphology AO (aortic valve opening) and AC (aortic valve closing) amplitudes showed significant decrease between BDC12 and HDT52 (p < 0.03). PTT and LVET were also found to decrease through HDT bed rest (p < 0.01). Furthermore, PTT and LVET magnitude of response to bed rest was found to be different between campaigns (p < 0.001) possibly due to seasonal effects on of the cardiovascular system. Correlations between FDA and cardiac timing intervals PTT and LVET using SCG suggests decreases in mechanical strength of the heart and increased arterial stiffness due to fluid shifts associated with the prolonged bed rest.

Detecting Coronary Artery Disease Using Rest Seismocardiography and Gyrocardiography.

In this study, we present a non-invasive solution to identify patients with coronary artery disease (CAD) defined as ⩾50% stenosis in at least one coronary artery. The solution is based on the analysis of linear acceleration (seismocardiogram, SCG) and angular velocity (gyrocardiogram, GCG) of the heart recorded in the x, y, and z directional axes from an accelerometer/gyroscope sensor mounted on the sternum. The database was collected from 310 individuals through a multicenter study. The time-frequency features extracted from each SCG and GCG data channel were fed to a one-dimensional Convolutional Neural Network (1D CNN) to train six separate classifiers. The results from different classifiers were later fused to estimate the CAD risk for each participant. The predicted CAD risk was validated against related results from angiography. The SCG z and SCG y classifiers showed better performance relative to the other models (p < 0.05) with the area under the curve (AUC) of 91%. The sensitivity range for CAD detection was 92-94% for the SCG models and 73-87% for the GCG models. Based on our findings, the SCG models achieved better performance in predicting the CAD risk compared to the GCG models; the model based on the combination of all SCG and GCG classifiers did not achieve higher performance relative to the other models. Moreover, these findings showed that the performance of the proposed 3-axial SCG/GCG solution based on recordings obtained during rest was comparable, or better than stress ECG. These data may indicate that 3-axial SCG/GCG could be used as a portable at-home CAD screening tool.

Investigating the estimation of cardiac time intervals using gyrocardiography.

OBJECTIVE: Assessment of cardiac time intervals (CTIs) is essential for monitoring cardiac performance. Recently, gyrocardiography (GCG) has been introduced as a non-invasive technology for cardiac monitoring. GCG measures the chest's angular precordial vibrations caused by myocardium wall motion using a gyroscope sensor attached to the sternum. In this study, we investigated the accuracy and reproducibility of estimating CTIs from the GCG recordings of 50 adults. APPROACH: We proposed five fiducial points for the GCG waveforms associated with the opening and closure of aortic and mitral valves. Two annotators annotated the suggested points on each cardiac cycle. The points were compared to the corresponding opening and closing of cardiac valves delineated on Tissue Doppler imaging (TDI) recordings. The fiducial points were annotated on seismocardiography (SCG) and impedance cardiography (ICG) signals recorded simultaneously. MAIN RESULTS: For estimating the timing of mitral valve closure, aortic valve opening, aortic valve closure, and mitral valve opening, 40%, 67%, 75%, and 70% of GCG annotations fell in the corresponding echocardiography ranges, respectively. The results showed moderate-to-excellent (r = 0.4-0.92; p-value < 0.01) correlation between the measured and the reference CTls. A myocardial performance index (Tei index) adapted using joint GCG and SCG resulted in a moderate correlation (r = 0.4; p-value < 0.001). SIGNIFICANCE: The findings showed that the CTIs can be easily measured using GCG. Also, we found that using SCG and GCG recordings together could provide an opportunity to estimate CTIs more accurately, and make it possible to calculate the Tei index as an indicator of myocardial performance.

The repeatability of estimated systolic time intervals in healthy subjects using seismocardiogram and electrocardiogram.

OBJECTIVE: We investigated the repeatability of systolic time intervals (STIs) in healthy subjects using a combination of seismocardiogram (SCG) and electrocardiogram (ECG). STIs have been extensively used in the past to quantify heart performance, particularly the left ventricle. In this study, STIs included pre-ejection period (PEP), left ventricular ejection time (LVET), and their ratio. APPROACH: We conducted the repeatability test of STI estimation through two experiments. The first involved three consecutive one-minute recordings separated by one-minute intervals, and the second involved two one-minute recordings separated by 24 h. Twenty healthy subjects participated in our study. We considered the coefficient of variation (CV) to quantify the repeatability. As there was no agreed upon values for optimal CV values, we compared our results with an alternative method using a combination of impedance cardiography (ICG) and ECG. Similar to our method, the alternative method was noninvasive and could be employed for personal heart monitoring. We also studied the repeatability after STIs were corrected for heart rate using two approaches. The first approach used a multiplicative factor per subject based on the heart rates in each recordings of that subject. The second approach employed sex-specific regression models for all subjects (Weissler's equations). MAIN RESULTS: We found that the repeatability of our method (SCG and ECG) was in agreement with the alternative method (ICG and ECG) in both experiments. Moreover, the Weissler's equations approach for heart rate increased the repeatability. SIGNIFICANCE: It can be concluded that estimation of PEP, LVET and their ratio through SCG and ECG signals was repeatable in healthy subjects.

Identifying Patients With Coronary Artery Disease Using Rest and Exercise Seismocardiography.

Coronary artery disease (CAD) is the most common cause of death globally. Patients with suspected CAD are usually assessed by exercise electrocardiography (ECG). Subsequent tests, such as coronary angiography and coronary computed tomography angiography (CCTA) are performed to localize the stenosis and to estimate the degree of blockage. The present study describes a non-invasive methodology to identify patients with CAD based on the analysis of both rest and exercise seismocardiography (SCG). SCG is a non-invasive technology for capturing the acceleration of the chest induced by myocardial motion and vibrations. SCG signals were recorded from 185 individuals at rest and immediately after exercise. Two models were developed using the characterization of the rest and exercise SCG signals to identify individuals with CAD. The models were validated against related results from angiography. For the rest model, accuracy was 74%, and sensitivity and specificity were estimated as 75 and 72%, respectively. For the exercise model accuracy, sensitivity, and specificity were 81, 82, and 84%, respectively. The rest and exercise models presented a bootstrap-corrected area under the curve of 0.77 and 0.91, respectively. The discrimination slope was estimated 0.32 for rest model and 0.47 for the exercise model. The difference between the discrimination slopes of these two models was 0.15 (95% CI: 0.10 to 0.23, p < 0.0001). Both rest and exercise models are able to detect CAD with comparable accuracy, sensitivity, and specificity. Performance of SCG is better compared to stress-ECG and it is identical to stress-echocardiography and CCTA. SCG examination is fast, inexpensive, and may even be carried out by laypersons.

Comparison of Different Methods for Estimating Cardiac Timings: A Comprehensive Multimodal Echocardiography Investigation.

Cardiac time intervals are important hemodynamic indices and provide information about left ventricular performance. Phonocardiography (PCG), impedance cardiography (ICG), and recently, seismocardiography (SCG) have been unobtrusive methods of choice for detection of cardiac time intervals and have potentials to be integrated into wearable devices. The main purpose of this study was to investigate the accuracy and precision of beat-to-beat extraction of cardiac timings from the PCG, ICG and SCG recordings in comparison to multimodal echocardiography (Doppler, TDI, and M-mode) as the gold clinical standard. Recordings were obtained from 86 healthy adults and in total 2,120 cardiac cycles were analyzed. For estimation of the pre-ejection period (PEP), 43% of ICG annotations fell in the corresponding echocardiography ranges while this was 86% for SCG. For estimation of the total systolic time (TST), these numbers were 43, 80, and 90% for ICG, PCG, and SCG, respectively. In summary, SCG and PCG signals provided an acceptable accuracy and precision in estimating cardiac timings, as compared to ICG.

Relationship between Ischemic Stroke and Pulse Rate Variability as a Surrogate of Heart Rate Variability.

Autonomic reflex ascertains cardiovascular homeostasis during standing. Impaired autonomic reflex could lead to dizziness and falls while standing; this is prevalent in stroke survivors. Pulse rate variability (PRV) has been utilized in the literature in lieu of heart rate variability (HRV) for ambulatory and portable monitoring of autonomic reflex predominantly in young, healthy individuals. Here, we compared the PRV with gold standard HRV for monitoring autonomic reflex in ischemic stroke survivors. Continuous blood pressure and electrocardiography were acquired from ischemic stroke survivors (64 ± 1 years) and age-matched controls (65 ± 2 years) during a 10-minute sit-to-stand test. Beat-by-beat heart period (represented by RR and peak-to-peak (PP) intervals), systolic blood pressure (SBP), diastolic blood pressure (DBP), and pulse arrival time (PAT), an indicator of arterial stiffness, were derived. Time and frequency domain HRV (from RR intervals) and PRV (from PP intervals) metrics were extracted. PAT was lower (248 ± 7 ms vs. 270 ± 8 ms, p < 0.05) suggesting higher arterial stiffness in stroke survivors compared to controls during standing. Further, compared to controls, the agreement between HRV and PRV was impaired in stroke survivors while standing. The study outcomes suggest that caution should be exercised when considering PRV as a surrogate of HRV for monitoring autonomic cardiovascular control while standing in stroke survivors.

Pediatric pulse oximetry-based OSA screening at different thresholds of the apnea-hypopnea index with an expression of uncertainty for inconclusive classifications.

BACKGROUND: Assessments of pediatric obstructive sleep apnea (OSA) are underutilized across Canada due to a lack of resources. Polysomnography (PSG) measures OSA severity through the average number of apnea/hypopnea events per hour (AHI), but is resource intensive and requires a specialized sleep laboratory, which results in long waitlists and delays in OSA detection. Prompt diagnosis and treatment of OSA are crucial for children, as untreated OSA is linked to behavioral deficits, growth failure, and negative cardiovascular consequences. We aim to assess the performance of a portable pediatric OSA screening tool at different AHI cut-offs using overnight smartphone-based pulse oximetry. MATERIAL AND METHODS: Following ethics approval and informed consent, children referred to British Columbia Children's Hospital for overnight PSG were recruited for two studies including 160 and 75 children, respectively. An additional smartphone-based pulse oximeter sensor was used in both studies to record overnight pulse oximetry [SpO2 and photoplethysmogram (PPG)] alongside the PSG. Features characterizing SpO2 dynamics and heart rate variability from pulse peak intervals of the PPG signal were derived from pulse oximetry recordings. Three multivariate logistic regression screening models, targeted at three different levels of OSA severity (AHI ≥ 1, 5, and 10), were developed using stepwise-selection of features using the Bayesian information criterion (BIC). The "Gray Zone" approach was also implemented for different tolerance values to allow for more precise detection of children with inconclusive classification results. RESULTS: The optimal diagnostic tolerance values defining the "Gray Zone" borders (15, 10, and 5, respectively) were selected to develop the final models to screen for children at AHI cut-offs of 1, 5, and 10. The final models evaluated through cross-validation showed good accuracy (75%, 82% and 89%), sensitivity (80%, 85% and 82%) and specificity (65%, 79% and 91%) values for detecting children with AHI ≥ 1, AHI ≥ 5 and AHI ≥ 10. The percentage of children classified as inconclusive was 28%, 38% and 16% for models detecting AHI ≥ 1, AHI ≥ 5, and AHI ≥ 10, respectively. CONCLUSIONS: The proposed pulse oximetry-based OSA screening tool at different AHI cut-offs may assist clinicians in identifying children at different OSA severity levels. Using this tool at home prior to PSG can help with optimizing the limited resources for PSG screening. Further validation with larger and more heterogeneous datasets is required before introducing in clinical practice.

Night to night pulse oximetry variability in children with suspected sleep apnea.

Obstructive Sleep Apnea (OSA) is the most common form of sleep-disordered breathing in children. The gold standard to screen for OSA, polysomnography (PSG), requires an overnight stay in the hospital and is resource intensive. The Phone Oximeter is a non-invasive smartphone-based tool to record pulse oximetry. This portable device is able to measure patients over multiple nights while at home, causing less sleep disturbance than PSG and is able to measure night to night variability in sleep. This study analyzed the Screen My Sleep children (SMS) dataset, in which 74 children were monitored over multiple nights with the Phone Oximeter, including one night simultaneously with PSG in the hospital and two nights at home. In this study, we aim to investigate the night to night variability and assess the accuracy of the oxygen desaturation index (ODI) screening for children with significant OSA. In order to assess the performance of the ODI calculation in children, we implemented different ODIs at different desaturation levels and time durations. The variability was studied using a one-way ANOVA, and ODI's performance screening for OSA using the area under the ROC curve (AUC). The implemented ODIs provide similar OSA screening results, using different apnea/hypopnea index (AHI) thresholds, as the ODI recommended for adults by the American academy of sleep medicine (AASM). The ODI provides an AUC of around 0.77, 0.76, 0.94 and 0.97 classifying children with an AHI > 1, AHI > 5 AHI > 10 and AHI > 15, respectively. The SMS dataset shows no significant night to night variability between the two nights at home. However, when comparing with the night at the hospital, both nights at home show a decrease in the lowest SpO2 value as well as overall SpO2 signal quality percentage. This study shows that there is variability in SpO2 signal between at-home versus in hospital settings.

Extracting Instantaneous Respiratory Rate From Multiple Photoplethysmogram Respiratory-Induced Variations.

In this study, we proposed a novel method for extracting the instantaneous respiratory rate (IRR) from the pulse oximeter photoplethysmogram (PPG). The method was performed in three main steps: (1) a time-frequency transform called synchrosqueezing transform (SST) was used to extract the respiratory-induced intensity, amplitude and frequency variation signals from PPG, (2) the second SST was applied to each extracted respiratory-induced variation signal to estimate the corresponding IRR, and (3) the proposed peak-conditioned fusion method then combined the IRR estimates to calculate the final IRR. We validated the implemented method with capnography and nasal/oral airflow as the reference RR using the limits of agreement (LOA) approach. Compared to simple fusion and single respiratory-induced variation estimations, peak-conditioned fusion shows better performance. It provided a bias of 0.28 bpm with the 95% LOAs ranging from -3.62 to 4.17, validated against capnography and a bias of 0.04 bpm with the 95% LOAs ranging from -5.74 to 5.82, validated against nasal/oral airflow. This algorithm would expand the functionality of a conventional pulse oximetry beyond the measurement of heart rate and oxygen saturation to measure the respiratory rate continuously and instantly.

Evaluation of cardiac modulation in children in response to apnea/hypopnea using the Phone Oximeter(™).

Individuals with sleep disordered breathing (SDB) can experience changes in automatic cardiac regulation as a result of frequent sleep fragmentation and disturbance in normal respiration and oxygenation that accompany most apnea/hypopnea events. In adults, these changes are reflected in enhanced sympathetic and reduced parasympathetic activity. In this study, we examined the autonomic cardiac regulation in children with and without SDB, through spectral and detrended fluctuation analysis (DFA) of pulse rate variability (PRV). PRV was measured from pulse-to-pulse intervals (PPIs) of the photoplethysmogram (PPG) recorded from 160 children using the Phone Oximeter(™) in the standard setting of overnight polysomnography. Spectral analysis of PRV showed the cardiac parasympathetic index (high frequency, HF) was lower (p < 0.01) and cardiac sympathetic indices (low frequency, LF and LF/HF ratio) were higher (p < 0.01) during apnea/hypopnea events for more than 95% of children with SDB. DFA showed the short- and long-range fluctuations of heart rate were more strongly correlated in children with SDB compared to children without SDB. These findings confirm that the analysis of the PPG recorded using the Phone Oximeter(™) could be the basis for a new screening tool for assessing PRV in non-clinical environment.

Estimating instantaneous respiratory rate from the photoplethysmogram.

The photoplethysmogram (PPG) obtained from pulse oximetry shows the local changes of blood volume in tissues. Respiration induces variation in the PPG baseline due to the variation in venous blood return during each breathing cycle. We have proposed an algorithm based on the synchrosqueezing transform (SST) to estimate instantaneous respiratory rate (IRR) from the PPG. The SST is a combination of wavelet analysis and a reallocation method which aims to sharpen the time-frequency representation of the signal and can provide an accurate estimation of instantaneous frequency. In this application, the SST was applied to the PPG and IRR was detected as the predominant ridge in the respiratory band (0.1 Hz - 1 Hz) in the SST plane. The algorithm was tested against the Capnobase benchmark dataset that contains PPG, capnography, and expert labelled reference respiratory rate from 42 subjects. The IRR estimation accuracy was assessed using the root mean square (RMS) error and Bland-Altman plot. The median RMS error was 0.39 breaths/min for all subjects which ranged from the lowest error of 0.18 breaths/min to the highest error of 13.86 breaths/min. A Bland-Altman plot showed an agreement between the IRR obtained from PPG and reference respiratory rate with a bias of -0.32 and limits agreement of -7.72 to 7.07. Extracting IRR from PPG expands the functionality of pulse oximeters and provides additional diagnostic power to this non-invasive monitoring tool.

Pulse oximetry recorded from the Phone Oximeter for detection of obstructive sleep apnea events with and without oxygen desaturation in children.

Obstructive sleep apnea (OSA) disrupts normal ventilation during sleep and can lead to serious health problems in children if left untreated. Polysomnography, the gold standard for OSA diagnosis, is resource intensive and requires a specialized laboratory. Thus, we proposed to use the Phone Oximeter™, a portable device integrating pulse oximetry with a smartphone, to detect OSA events. As a proportion of OSA events occur without oxygen desaturation (defined as SpO2 decreases ≥ 3%), we suggest combining SpO2 and pulse rate variability (PRV) analysis to identify all OSA events and provide a more detailed sleep analysis. We recruited 160 children and recorded pulse oximetry consisting of SpO2 and plethysmography (PPG) using the Phone Oximeter™, alongside standard polysomnography. A sleep technician visually scored all OSA events with and without oxygen desaturation from polysomnography. We divided pulse oximetry signals into 1-min signal segments and extracted several features from SpO2 and PPG analysis in the time and frequency domain. Segments with OSA, especially the ones with oxygen desaturation, presented greater SpO2 variability and modulation reflected in the spectral domain than segments without OSA. Segments with OSA also showed higher heart rate and sympathetic activity through the PRV analysis relative to segments without OSA. PRV analysis was more sensitive than SpO2 analysis for identification of OSA events without oxygen desaturation. Combining SpO2 and PRV analysis enhanced OSA event detection through a multiple logistic regression model. The area under the ROC curve increased from 81% to 87%. Thus, the Phone Oximeter™ might be useful to monitor sleep and identify OSA events with and without oxygen desaturation at home.

Development of a screening tool for sleep disordered breathing in children using the phone Oximeter™.

BACKGROUND: Sleep disordered breathing (SDB) can lead to daytime sleepiness, growth failure and developmental delay in children. Polysomnography (PSG), the gold standard to diagnose SDB, is a highly resource-intensive test, confined to the sleep laboratory. AIM: To combine the blood oxygen saturation (SpO2) characterization and cardiac modulation, quantified by pulse rate variability (PRV), to identify children with SDB using the Phone Oximeter, a device integrating a pulse oximeter with a smartphone. METHODS: Following ethics approval and informed consent, 160 children referred to British Columbia Children's Hospital for overnight PSG were recruited. A second pulse oximeter sensor applied to the finger adjacent to the one used for standard PSG was attached to the Phone Oximeter to record overnight pulse oximetry (SpO2 and photoplethysmogram (PPG)) alongside the PSG. RESULTS: We studied 146 children through the analysis of the SpO2 pattern, and PRV as an estimate of heart rate variability calculated from the PPG. SpO2 variability and SpO2 spectral power at low frequency, was significantly higher in children with SDB due to the modulation provoked by airway obstruction during sleep (p-value <0.01). PRV analysis reflected a significant augmentation of sympathetic activity provoked by intermittent hypoxia in SDB children. A linear classifier was trained with the most discriminating features to identify children with SDB. The classifier was validated with internal and external cross-validation, providing a high negative predictive value (92.6%) and a good balance between sensitivity (88.4%) and specificity (83.6%). Combining SpO2 and PRV analysis improved the classification performance, providing an area under the receiver operating characteristic curve of 88%, beyond the 82% achieved using SpO2 analysis alone. CONCLUSIONS: These results demonstrate that the implementation of this algorithm in the Phone Oximeter will provide an improved portable, at-home screening tool, with the capability of monitoring patients over multiple nights.

Assessment of respiratory flow and efforts using upper-body acceleration.

In this paper, an innovative method for estimating the respiratory flow and efforts is proposed and evaluated in various postures and flow rates. Three micro electro-mechanical system accelerometers were mounted on the suprasternal notch, thorax and abdomen of subjects in supine, prone and lateral positions to record the upper airway acceleration and the movements of the chest and abdomen wall. The respiratory flow and efforts were estimated from the recorded acceleration signals by applying machine learning methods. To assess the agreement of the estimated signals with the well-established measurement methods, standard error of measurement (SEM) was calculated and ρ=1-SEM was estimated for every condition. A significant agreement between the estimated and reference signals was found (ρ=0.83, 0.82 and 0.89 for the estimated flow, thorax and abdomen efforts respectively). Additionally, the agreement of the estimated and reference flows was assessed by calculating the ratio of time at the tidal peak inspiration flow to the inspiration time (tPTIF/tI) and the ratio of time at the tidal peak expiration flow to the expiration time (tPTEF/tE). Overall mean and standard deviation of absolute value of differences between tPTIF/tI and tPTEF/tE ratios calculated for every breathing cycle of reference and estimated flow were 0.0035 (0.06) and 0.051 (0.032), respectively. The presented results demonstrate the feasibility of using the upper-body acceleration as a simple solution for long-term monitoring of respiratory features.

Oxygen saturation resolution influences regularity measurements.

The measurement of regularity in the oxygen saturation (SpO(2)) signal has been suggested for use in identifying subjects with sleep disordered breathing (SDB). Previous work has shown that children with SDB have lower SpO(2) regularity than subjects without SDB (NonSDB). Regularity was measured using non-linear methods like approximate entropy (ApEn), sample entropy (SamEn) and Lempel-Ziv (LZ) complexity. Different manufacturer's pulse oximeters provide SpO(2) at various resolutions and the effect of this resolution difference on SpO(2) regularity, has not been studied. To investigate this effect, we used the SpO(2) signal of children with and without SDB, recorded from the Phone Oximeter (0.1% resolution) and the same SpO(2) signal rounded to the nearest integer (artificial 1% resolution). To further validate the effect of rounding, we also used the SpO(2) signal (1% resolution) recorded simultaneously from polysomnography (PSG), as a control signal. We estimated SpO(2) regularity by computing the ApEn, SamEn and LZ complexity, using a 5-min sliding window and showed that different resolutions provided significantly different results. The regularity calculated using 0.1% SpO(2) resolution provided no significant differences between SDB and NonSDB. However, the artificial 1% resolution SpO(2) provided significant differences between SDB and NonSDB, showing a more random SpO(2) pattern (lower SpO(2) regularity) in SDB children, as suggested in the past. Similar results were obtained with the SpO(2) recorded from PSG (1% resolution), which further validated that this SpO(2) regularity change was due to the rounding effect. Therefore, the SpO(2) resolution has a great influence in regularity measurements like ApEn, SamEn and LZ complexity that should be considered when studying the SpO(2) pattern in children with SDB.

Pulse rate variability compared with Heart Rate Variability in children with and without sleep disordered breathing.

Heart Rate Variability (HRV), the variation of time intervals between heartbeats, is one of the most promising and widely used quantitative markers of autonomic activity. Traditionally, HRV is measured as the series of instantaneous cycle intervals obtained from the electrocardiogram (ECG). In this study, we investigated the estimation of variation in heart rate from a photoplethysmography (PPG) signal, called pulse rate variability (PRV), and assessed its accuracy as an estimate of HRV in children with and without sleep disordered breathing (SDB). We recorded raw PPGs from 72 children using the Phone Oximeter, an oximeter connected to a mobile phone. Full polysomnography including ECG was simultaneously recorded for each subject. We used correlation and Bland-Altman analysis for comparing the parameters of HRV and PRV between two groups of children. Significant correlation (r > 0.90, p < 0.05) and close agreement were found between HRV and PRV for mean intervals, standard deviation of intervals (SDNN) and the root-mean square of the difference of successive intervals (RMSSD). However Bland-Altman analysis showed a large divergence for LF/HF ratio parameter. In addition, children with SDB had depressed SDNN and RMSSD and elevated LF/HF in comparison to children without SDB. In conclusion, PRV provides the accurate estimate of HRV in time domain analysis but does not reflect precise estimation for parameters in frequency domain.

Oxygen saturation in children with and without obstructive sleep apnea using the phone-oximeter.

Obstructive sleep apnea (OSA) in children can lead to daytime sleepiness, growth failure and developmental delay. Polysomnography (PSG), the gold standard to diagnose OSA is highly resource intensive and is confined to the sleep laboratory. In this study we propose to identify children with OSA using blood oxygen saturation (SpO2) obtained from the Phone Oximeter. This portable, in-home device is able to monitor patients over multiple nights, causes less sleep disturbance and facilitates a more natural sleep pattern. The proposed algorithm analyzes the SpO2 signal in the time and frequency domain using a 90-s sliding window. Three spectral parameters are calculated from the power spectral density (PSD) to evaluate the modulation in the SpO2 due to the oxyhemoblobin desaturations. The power P, slope S in the discriminant band (DB), and ratio R between P and total power are calculated for each window. Tendency and variability indices, number of SpO2 desaturations and time spent under 2% or 3% of baseline saturation level are computed for each time window. The statistical distribution of the temporal evolution of all parameters is analyzed to identify 68 children, 30 with OSA and 38 without OSA (nonOSA). This characterization was evaluated by a feature selection based on a linear discriminant. The combination of temporal and spectral parameters provided the best leave one out crossvalidation results with an accuracy of 86.8%, a sensitivity of 80.0%, and a specificity of 92.1% using only 5 parameters. The median of R, mean of P and S and mean and standard deviation of the number of desaturations below 3% of baseline saturation level, were the most representative parameters. Hence, a better knowledge of SpO2 dynamics could help identifying children with OSA with the Phone Oximeter.

Monitoring torso acceleration for estimating the respiratory flow and efforts for sleep apnea detection.

Sleep apnea syndrome is a common sleep breathing disorder classified into two major categories: obstructive and central. In this study, we propose a method based on ensemble learning to estimate the respiratory flow, the thoracic respiratory effort and the abdominal respiratory effort from acceleration of suprasternal notch, the thorax and the abdomen respectively. The estimated flow can be used to detect the breathing cessations and the estimated efforts can be used to classify them into obstructive and central apneas. The estimated signals are compared with the signals recorded by well-established measurement methods to show overall mean errors from 11% for the abdomen effort, 17% for the thorax effort and 16% error for the flow estimation. The presented results demonstrate the feasibility of using the torso acceleration as a simple and inexpensive solution for long term measuring and monitoring of respiratory functions for sleep apnea detection.

Validation of respiratory signal derived from suprasternal notch acceleration for sleep apnea detection.

This study evaluates the respiration signal derived from an accelerometer mounted on the suprasternal notch in three body positions and three respiration types simulating normal sleep conditions. The Acceleration Derived Respiratory signal (ADR) is compared with single strain gauge belt and a standard spirometry signal taken as reference. The results demonstrate the potential of ADR as a simple, low cost and unintrusive method of screening breath disorders such as obstructive sleep apnea/hypopnea.