Martin O. Mendez

Sleep Staging Based on Signals Acquired Through Bed Sensor

We describe a system for the evaluation of the sleep macrostructure on the basis of Emfit sensor foils placed into bed mattress and of advanced signal processing. The signals on which the analysis is based are heart-beat interval (HBI) and movement activity obtained from the bed sensor, the relevant features and parameters obtained through a time-variant autoregressive model (TVAM) used as feature extractor, and the classification obtained through a hidden Markov model (HMM). Parameters coming from the joint probability of the HBI features were used as input to a HMM, while movement features are used for wake period detection. A total of 18 recordings from healthy subjects, including also reference polysomnography, were used for the validation of the system. When compared to wake-nonrapid-eye-movement (NREM)-REM classification provided by experts, the described system achieved a total accuracy of 79±9% and a kappa index of 0.43±0.17 with only two HBI features and one movement parameter, and a total accuracy of 79±10% and a kappa index of 0.44±0.19 with three HBI features and one movement parameter. These results suggest that the combination of HBI and movement features could be a suitable alternative for sleep staging with the advantage of low cost and simplicity.

Sleep Apnea Screening by Autoregressive Models From a Single ECG Lead

This paper presents a method for obstructive sleep apnea (OSA) screening based on the electrocardiogram (ECG) recording during sleep. OSA is a common sleep disorder produced by repetitive occlusions in the upper airways and this phenomenon can usually be observed also in other peripheral systems such as the cardiovascular system. Then the extraction of ECG characteristics, such as the RR intervals and the area of the QRS complex, is useful to evaluate the sleep apnea in noninvasive way. In the presented analysis, 50 recordings coming from the apnea Physionet database were used; data were split into two sets, the training and the testing set, each of which was composed of 25 recordings. A bivariate time-varying autoregressive model (TVAM) was used to evaluate beat-by-beat power spectral densities for both the RR intervals and the QRS complex areas. Temporal and spectral features were changed on a minute-by-minute basis since apnea annotations where given with this resolution. The training set consisted of 4950 apneic and 7127 nonapneic minutes while the testing set had 4428 apneic and 7927 nonapneic minutes. The K-nearest neighbor (KNN) and neural networks (NN) supervised learning classifiers were employed to classify apnea and non apnea minutes. A sequential forward selection was used to select the best feature subset in a wrapper setting. With ten features the KNN algorithm reached an accuracy of 88%, sensitivity equal to 85%, and specificity up to 90%, while NN reached accuracy equal to 88%, sensitivity equal to 89% and specificity equal to 86%. In addition to the minute-by-minute classification, the results showed that the two classifiers are able to separate entirely (100%) the normal recordings from the apneic recordings. Finally, an additional database with eight recordings annotated as normal or apneic was used to test again the classifiers. Also in this new dataset, the results showed a complete separation between apneic and normal recordings.

Automatic screening of obstructive sleep apnea from the ECG based on empirical mode decomposition and wavelet analysis.

This study analyses two different methods to detect obstructive sleep apnea (OSA) during sleep time based only on the ECG signal. OSA is a common sleep disorder caused by repetitive occlusions of the upper airways, which produces a characteristic pattern on the ECG. ECG features, such as the heart rate variability (HRV) and the QRS peak area, contain information suitable for making a fast, non-invasive and simple screening of sleep apnea. Fifty recordings freely available on Physionet have been included in this analysis, subdivided in a training and in a testing set. We investigated the possibility of using the recently proposed method of empirical mode decomposition (EMD) for this application, comparing the results with the ones obtained through the well-established wavelet analysis (WA). By these decomposition techniques, several features have been extracted from the ECG signal and complemented with a series of standard HRV time domain measures. The best performing feature subset, selected through a sequential feature selection (SFS) method, was used as the input of linear and quadratic discriminant classifiers. In this way we were able to classify the signals on a minute-by-minute basis as apneic or nonapneic with different best-subset sizes, obtaining an accuracy up to 89% with WA and 85% with EMD. Furthermore, 100% correct discrimination of apneic patients from normal subjects was achieved independently of the feature extractor. Finally, the same procedure was repeated by pooling features from standard HRV time domain, EMD and WA together in order to investigate if the two decomposition techniques could provide complementary features. The obtained accuracy was 89%, similarly to the one achieved using only Wavelet analysis as the feature extractor; however, some complementary features in EMD and WA are evident.

Sleep staging from Heart Rate Variability: time-varying spectral features and Hidden Markov Models

An alternative DSS which models the behaviour of the Heart Rate Variability (HRV) signal linked to stable (NREM) and instable (REM) cerebral waves during sleep and a probabilistic model of the sleep stages transitions for decision was developed. Time-Varying Autoregressive Models (TVAMs) were used as feature extractor while Hidden Markov Models (HMM) was used as time series classifier. 24 full polysomnography recordings from healthy sleepers were used for the analysis and those were separated in two sets of

Detection of Sleep Apnea from surface ECG based on features extracted by an Autoregressive Model

This study proposes an alternative evaluation of obstructive sleep apnea (OSA) based on ECG signal during sleep time. OSA is a common sleep disorder produced by repetitive occlusions in the upper airways. This respiratory disturbance produces a specific pattern on ECG. Extraction of ECG characteristics, as heart rate variability (HRV) and peak R area, offers alternative measures for a sleep apnea pre-diagnosis. 50 recordings coming from the apnea Physionet database were used in the analysis, this database is part of the 70 recordings used for the Computer in Cardiology challenge celebrated in 2000. A bivariate autoregressive model was used to evaluate beat-by-beat power spectral density of HRV and R peak area. K-nearest neighbor (KNN) supervised learning classifier was employed for categorizing apnea events from normal ones, on a minute-by-minute basis for each recording. Data were split into two sets, training and testing set, each one with 25 recordings. The classification results showed an accuracy higher than 85% in both training and testing. In addition it was possible to separate completely between apnea and normal subjects and almost completely among apnea, normal and borderline subjects.

Discrimination of Sleep-Apnea-Related Decreases in the Amplitude Fluctuations of PPG Signal in Children by HRV Analysis

In this paper, an analysis of heart rate variability (HRV) during decreases in the amplitude fluctuations of photopletysmography (PPG) [decreases in the amplitude fluctuations of photopletysmography (DAP)] events for obstructive sleep apnea syndrome (OSAS) screening is presented. Two hundred and sixty-eight selected signal segments around the DAP event were extracted and classified in five groups depending on SaO2 and respiratory behavior. Four windows around each DAP are defined and temporal evolution of time-frequency HRV parameters was analyzed for OSAS screening. Results show a significant increase in sympathetic activity during DAP events, which is higher in cases associated with apnea. DAP events were classified as apneic or nonapneic using a linear discriminant analysis from the HRV indexes. The ratio of DAP events per hour r DAP and the ratio of apneic DAP events per hour r alpha DAP were computed. Results show an accuracy of 79% for r alpha DAP (12% increase with respect to r DAP), a sensitivity of 87.5%, and a specificity of 71.4% when classifying 1-h polysomnographic excerpts. As for clinical subject classification, an accuracy of 80% (improvement of 6.7% ), a sensitivity of 87.5%, and a specificity of 71.4% are reached. These results suggest that the combination of DAP and HRV could be an improved alternative for sleep apnea screening from PPG with the added benefit of its low cost and simplicity.

On Arousal from Sleep: Time-Frequency Analysis

Time-frequency analysis of the heart rate variability during arousal from sleep, with and without EMG activation, coming from five obese healthy subjects was performed. Additionally, a comparative analysis of three time-frequency distributions, smooth pseudo Wigner–Ville (SPWVD), Choi–Williams (CWD) and Born–Jordan distribution (BJD) is presented in this study. SPWVD showed higher capacity for eliminating the cross terms independently of the signal. After applying Hilbert transformation to real signals BJD and CWD lost some important mathematic properties as marginals, on the contrary PSWVD remains unchanged. BJD showed results comparable with CWD. During arousal episodes, analogous energy distribution and spectral indexes were obtained by the three time-frequency representations. Arousals with chin activity presented stronger changes in RR intervals and LF (related to sympathetic activity) component, being statistically different with respect to arousal without chin activity, only around the period of maximum change in β activity on the EEG. These results suggest a more evident stress for the heart when an arousal is related to external muscular activity.

Processing of Signals Recorded Through Smart Devices: Sleep-Quality Assessment

In this paper, we discuss the possibility of performing a sleep evaluation from signals, which are not usually used for this purpose. In particular, we take into consideration the heart rate variability (HRV) and respiratory signals for automatic sleep staging, arousals detection, and apnea recognition. This is particularly useful for wearable or textile devices that could be employed for home monitoring of sleep. The HRV and the respiration were analyzed in the frequency domain, and the statistics on the spectral and cross-spectral parameters put into evidence the possibility of a sleep evaluation on their basis. Comparison with traditional polysomnography (PSG) revealed a classification accuracy of 89.9% in rapid eye movement (REM) non-REM sleep separation and an accuracy of 88% for sleep apnea detection. Additional information can be achieved from the number of microarousals recognized in correspondence of typical modifications in the HRV signal. The obtained results support the idea of automatic sleep evaluation and monitoring through signals that are not traditionally used in clinical PSG, but can be easily recorded at home through wearable devices (for example, a sensorized T-shirt) or systems integrated into the environment (a sensorized bed). This is a first step for the development of systems for sleep screening on large populations that can constitute a complement for the traditional clinical evaluation.

Automatic sleep staging based on ballistocardiographic signals recorded through bed sensors

This study presents different methods for automatic sleep classification based on heart rate variability (HRV), respiration and movement signals recorded through bed sensors. Two methods for feature extraction have been implemented: time variant-autoregressive model (TVAM) and wavelet discrete transform (WDT); the obtained features are fed into two classifiers: Quadratic (QD) and Linear (LD) discriminant for staging sleep in REM, nonREM and WAKE periods. The performances of all the possible combinations of feature extractors and classifiers are compared in terms of accuracy and kappa index, using clinical polysomographyc evaluation as golden standard. 17 recordings from healthy subjects, including also polisomnography, were used to train and test the algorithms. When automatic classification is compared. QD-TVAM algorithm achieved a total accuracy of 76.81 ± 7.51 % and kappa index of 0.55 ± 0.10, while LD-WDT achieved a total accuracy of 79 ± 10% and kappa index of 0.51 ± 0.17. The results suggest that a good sleep evaluation can be achieved through non-conventional recording systems that could be used outside sleep centers.

Sleep Monitoring Through a Textile Recording System

In this paper, we present a home device for the continuous monitoring of sleep and investigate its reliability regarding sleep evaluation. The system has been particularly designed for healthy people and for preventive purposes. It is not obtrusive and therefore can be used every night without impeding sleep in itself and without interfering with the normal way of life. The signal used for sleep evaluation is the HRV derived from the ECG recorded by means of a sheet and a pillow. Patients in a sleep lab and healthy subjects at home were monitored during sleep with the textile system, while also standard ECG and respiration were recorded. For the textile ECG sensor, coverage of the signal on a beat-to-beat basis ranged from 47,9 - 95,8% of the overall night for the healthy subjects, with a mean coverage of 81,8%. In the group of sleep laboratory patients, the mean coverage was lower - 64,4% - although even in this group the coverage of a single night ranged up to 98.4%. After frequency analysis, the spectral parameters used for sleep staging and derived at the same time from standard and textile ECG signals were compared. The trends along the night are very similar, indicating the possibility of using textile HRV for sleep evaluation.