Oximetry Signals Research Articles

Deep-Learning Model Based on Convolutional Neural Networks to Classify Apnea-Hypopnea Events from the Oximetry Signal.

Automated analysis of the blood oxygen saturation (SpO2) signal from nocturnal oximetry has shown usefulness to simplify the diagnosis of obstructive sleep apnea (OSA), including the detection of respiratory events. However, the few preceding studies using SpO2 recordings have focused on the automated detection of respiratory events versus normal respiration, without making any distinction between apneas and hypopneas. In this sense, the characteristics of oxygen desaturations differ between obstructive apnea and hypopnea episodes. In this chapter, we use the SpO2 signal along with a convolutional neural network (CNN)-based deep-learning architecture for the automatic identification of apnea and hypopnea events. A total of 398 SpO2 signals from adult OSA patients were used for this purpose. A CNN architecture was trained using 30-s epochs from the SpO2 signal for the automatic classification of three classes: normal respiration, apnea, and hypopnea. Then, the apnea index (AI), the hypopnea index (HI), and the apnea-hypopnea index (AHI) were obtained by aggregating the outputs of the CNN for each subject (AICNN, HICNN, and AHICNN). This model showed a promising diagnostic performance in an independent test set, with 80.3% 3-class accuracy and 0.539 3-class Cohen's kappa for the classification of respiratory events. Furthermore, AICNN, HICNN, and AHICNN showed a high agreement with the values obtained from the standard PSG: 0.8023, 0.6774, and 0.8466 intra-class correlation coefficients (ICCs), respectively. This suggests that CNN can be used to analyze SpO2 recordings for the automated diagnosis of OSA in at-home oximetry tests.

Oximetry Indices in the Management of Sleep Apnea: From Overnight Minimum Saturation to the Novel Hypoxemia Measures.

Obstructive sleep apnea (OSA) is a multidimensional disease often underdiagnosed due to the complexity and unavailability of its standard diagnostic method: the polysomnography. Among the alternative abbreviated tests searching for a compromise between simplicity and accurateness, oximetry is probably the most popular. The blood oxygen saturation (SpO2) signal is characterized by a near-constant profile in healthy subjects breathing normally, while marked drops (desaturations) are linked to respiratory events. Parameterization of the desaturations has led to a great number of indices of severity assessment commonly used to assist in OSA diagnosis. In this chapter, the main methodologies used to characterize the overnight oximetry profile are reviewed, from visual inspection and simple statistics to complex measures involving signal processing and pattern recognition techniques. We focus on the individual performance of each approach, but also on the complementarity among the great amount of indices existing in the state of the art, looking for the most relevant oximetric feature subset. Finally, a quick overview of SpO2-based deep learning applications for OSA management is carried out, where the raw oximetry signal is analyzed without previous parameterization. Our research allows us to conclude that all the methodologies (conventional, time, frequency, nonlinear, and hypoxemia-based) demonstrate high ability to provide relevant oximetric indices, but only a reduced set provide non-redundant complementary information leading to a significant performance increase. Finally, although oximetry is a robust tool, greater standardization and prospective validation of the measures derived from complex signal processing techniques are still needed to homogenize interpretation and increase generalizability.

Influence of Melanin Content on Laser Doppler Flowmetry and Tissue Reflectance Oximetry Signal Formation

The article presents the results of studies of the melanin effect on recorded signals in laser Doppler flowmetry (LDF) and tissue reflectance oximetry (TO). By investigating the properties of recorded microvascular blood flow and skin oxygenation signals, we have gained new knowledge about race-related differences in the formation of these signals. The effect of the skin melanin content on the intensity of the recorded signals in the LDF and TO methods was evaluated, and the limitations in the use of these methods in representatives of different ethnic groups were shown.

Influence of melanin content on laser Doppler flowmetry and tissue reflectance oximetry signal formation

Influence of melanin content on laser Doppler flowmetry and tissue reflectance oximetry signal formation

Predicting Cardiovascular Outcomes Using the Respiratory Event Desaturation Transient Area Derived from Overnight Sleep Studies.

A new method for calculation of an overnight oximetry signal metric which is predictive of cardiovascular disease (CVD) outcomes in individuals undergoing an overnight sleep test is presented. The metric - the respiratory event desaturation transient area (REDTA) - quantifies the desaturation associated with respiratory events. Data from the Sleep Heart Health Study, which includes overnight oximetry signals and long-term CVD outcomes, was used to develop and test the parameter. Performance of the REDTA parameter was assessed using Cox proportional hazard ratios and compared to established metrics of hypoxia. Results show that hazard ratios in adjusted Cox analysis for predicting cardiovascular death using REDTA are up to 1.90 (95%CI: 1.22-2.96) which compares with the best of the established metrics. A big advantage of our metric compared to other high performing metrics is its ease of computation.

Automated Sleep apnea detection using optimal duration-frequency concentrated wavelet-based features of pulse oximetry signals

Automated Sleep apnea detection using optimal duration-frequency concentrated wavelet-based features of pulse oximetry signals

Clinical Applications of Near-infrared Spectroscopy Monitoring in Cardiovascular Surgery.

Near-infrared spectroscopy monitoring provides a practical method to follow trends in superficial cerebral cortex oxygenation during and after cardiovascular surgery. Determination of the limits of cerebral pressure–flow autoregulation is now possible using processed oximetry signals in relation to arterial pressure.

A novel algorithm for automatic diagnosis of sleep apnea from airflow and oximetry signals

Objective. Sleep apnea significantly decreases the quality of life. The apnea hypopnea index (AHI) is the main indicator for sleep apnea diagnosis. This study explored a novel automatic algorithm to diagnose sleep apnea from nasal airflow (AF) and pulse oximetry (SpO2) signals. Approach. Of the 988 polysomnography (PSG) records from the sleep heart health study (SHHS), 45 were randomly selected for the development of an algorithm and the remainder for validation (n = 943). The algorithm detects apnea events by a digitization process, following the determination of the peak excursion (peak-to-trough amplitude) from AF envelope. Hypopnea events were determined from the AF envelope and oxygen desaturation with correction to time lag in SpO2. Total sleep time (TST) was estimated from an optimized percentage of artefact-free total recording time. AHI was estimated from the number of detected events divided by the estimated TST. The estimated AHI was compared to the scored SHHS data for performance evaluation. Main results. The validation showed good agreement between the estimated and scored AHI (intraclass correlation coefficient of 0.95 and mean ±95% limits of agreement of −1.6 ±12.5 events h−1). The diagnostic accuracies were found: 90.7%, 91%, and 96.7% for AHI cut-off ≥5, ≥15, and ≥30 respectively. Significance. The new algorithm is accurate over other existing methods for the automatic diagnosis of sleep apnea. It is applicable to any portable sleep screeners especially for the home diagnosis of sleep apnea.

Classifying sleep–wake stages through recurrent neural networks using pulse oximetry signals

The regulation of the autonomic nervous system changes with the sleep stages causing variations in the physiological variables. We exploit these changes with the aim of classifying the sleep stages in awake or asleep using pulse oximeter signals. We applied a recurrent neural network to heart rate and peripheral oxygen saturation signals to classify the sleep stage every 30 seconds. The network architecture consists of two stacked layers of bidirectional gated recurrent units (GRUs) and a softmax layer to classify the output. In this paper, we used 5000 patients from the Sleep Heart Health Study dataset. 2500 patients were used to train the network, and two subsets of 1250 were used to validate and test the trained models. In the test stage, the best result obtained was 90.13% accuracy, 94.13% sensitivity, 80.26% specificity, 92.05% precision, and 84.68% negative predictive value. Further, the Cohen’s Kappa coefficient was 0.74 and the average absolute error percentage to the actual sleep time was 8.9%. The performance of the proposed network is comparable with the state-of-the-art algorithms when they use much more informative signals (except those with EEG).

Power spectral densities of nocturnal pulse oximetry signals differ in OSA patients with and without daytime sleepiness

BackgroundAs nocturnal hypoxemia and heart rate variability are associated with excessive daytime sleepiness (EDS) related to OSA, we hypothesize that the power spectral densities (PSD) of nocturnal pulse oximetry signals could be utilized in the assessment of EDS. Thus, we aimed to investigate if PSDs contain features that are related to EDS and whether a convolutional neural network (CNN) could detect patients with EDS using self-learned PSD features. MethodsA total of 915 OSA patients who had undergone polysomnography with multiple sleep latency test on the following day were investigated. PSDs for nocturnal blood oxygen saturation (SpO2), heart rate (HR), and photoplethysmogram (PPG), as well as power in the 15–35 mHz band in SpO2 (PSPO2) and HR (PHR), were computed. Differences in PSD features were investigated between EDS groups. Additionally, a CNN classifier was developed for identifying severe EDS patients based on spectral data. ResultsSpO2 power content increased significantly (p < 0.002) with increasing severity of EDS. Furthermore, a significant (p < 0.001) increase in HR-PSD was found in severe EDS (mean sleep latency < 5 min). Elevated odds of having severe EDS was found in PSPO2 (OR = 1.19–1.29) and PHR (OR = 1.81–1.83). Despite these significant spectral differences, the CNN classifier reached only moderate sensitivity (49.5%) alongside high specificity (80.4%) in identifying patients with severe EDS. ConclusionsWe conclude that PSDs of nocturnal pulse oximetry signals contain features significantly associated with OSA-related EDS. However, CNN-based identification of patients with EDS is challenging via pulse oximetry.

Automatic Assessment of Pediatric Sleep Apnea Severity Using Overnight Oximetry and Convolutional Neural Networks.

In this study, we use the overnight blood oxygen saturation (SpO2) signal along with convolutional neural networks (CNN) for the automatic estimation of pediatric sleep apnea-hypopnea syndrome (SAHS) severity. The few preceding studies have focused on the application of conventional feature extraction methods to obtain information from the SpO2 signal, which may omit relevant data related to the illness. In contrast, deep learning techniques are able to automatically learn features from raw input signal. Thus, we propose to assess whether CNN, a deep learning algorithm, could automatically estimate the apnea-hypopnea index (AHÍ) from nocturnal oximetry to help establish pediatric SAHS presence and severity. A database of 746 SpO2 recordings is involved in the study. CNN was trained using 20-min segments from the SpO2 signal in the training set (400 subjects). Hyperparameters of the CNN architecture were tuned using a validation set (100 subjects). This model was applied to a test set (246 subjects), in which the final AHI of each patient was obtained as the average of the output of the CNN for all the segments of the corresponding SpO2 signal. The AHI estimated by the CNN showed a promising diagnostic performance, with 74.8%, 90.7%, and 95.1% accuracies for the common AHI severity thresholds of 1, 5, and 10 events per hour (e/h), respectively. Furthermore, this model reached 28.6, 32.9, and 120.0 positive likelihood ratios for the above-mentioned AHI thresholds. This suggests that the information extracted from the oximetry signal by deep learning techniques may be useful to both establish pediatric SAHS and its severity.

Is Cumulative Time of Oxygen Desaturation a Better Predictor of Cardiovascular Mortality than Apnoea Hypopnoea Index?

In this paper, we explored the link between sleep apnoea and cardiovascular disease (CVD) using a time-series statistical measure of sleep apnoea-related oxygen desaturation. We compared the performance of a hypoxic measure derived from the polysomnogram with the Apnoea Hypopnoea Index (AHI) in predicting CVD mortality in patients of the Sleep Heart Health Study.We estimated the relative cumulative time of SpO2 below 90% (Tr90) using pulse oximetry signals from polysomnogram recordings as the hypoxic measure of desaturation patterns. Then, the survival curves for hypoxia quintiles were evaluated for the prediction of CVD mortality and were compared with the results using AHI for prediction. We also calculated the Cox hazard ratios for Tr90 and AHI. Our results show that the Tr90 was a better predictor of CVD mortality outcomes than AHI.

A Per-sample Digitized Algorithm for Automatically Detecting Apnea and Hypopnea Events from Airflow and Oximetry.

Sleep apnea is a common sleep disorder that can significantly decrease the quality of life. An accurate and early diagnosis of sleep apnea is required before getting proper treatment. A reliable automated detection of sleep apnea can overcome the problems of manual diagnosis (scoring) due to variability in recording and scoring criteria (for example across Europe) and to inter-scorer variability. This study explored a novel automated algorithm to detect apnea and hypopnea events from airflow and pulse oximetry signals, extracted from 30 polysomnography records of the Sleep Heart Health Study. Apneas and hypopneas were manually scored by a trained sleep physiologist according to the updated 2017 American Academy of Sleep Medicine respiratory scoring rules. From pre-processed airflow, the peak signal excursion was precisely determined from the peak-to-trough amplitude using a sliding window, with a per-sample digitized algorithm for detecting apnea and hypopnea. For apnea, the peak signal excursion drop was operationalized at ≥85% and for hypopnea at ≥35% of its pre-event baseline. Using backward shifting of oximetry, hypopneas were filtered with ≥3% oxygen desaturation from its baseline. The performance of the automated algorithm was evaluated by comparing the detection with manual scoring (a standard practice). The sensitivity and positive predictive value of detecting apneas and hypopneas were respectively 98.1% and 95.3%. This automated algorithm is applicable to any portable sleep monitoring device for the accurate detection of sleep apnea.

Assessment of Airflow and Oximetry Signals to Detect Pediatric Sleep Apnea-Hypopnea Syndrome Using AdaBoost.

The reference standard to diagnose pediatric Obstructive Sleep Apnea (OSA) syndrome is an overnight polysomnographic evaluation. When polysomnography is either unavailable or has limited availability, OSA screening may comprise the automatic analysis of a minimum number of signals. The primary objective of this study was to evaluate the complementarity of airflow (AF) and oximetry (SpO2) signals to automatically detect pediatric OSA. Additionally, a secondary goal was to assess the utility of a multiclass AdaBoost classifier to predict OSA severity in children. We extracted the same features from AF and SpO2 signals from 974 pediatric subjects. We also obtained the 3% Oxygen Desaturation Index (ODI) as a common clinically used variable. Then, feature selection was conducted using the Fast Correlation-Based Filter method and AdaBoost classifiers were evaluated. Models combining ODI 3% and AF features outperformed the diagnostic performance of each signal alone, reaching 0.39 Cohens’s kappa in the four-class classification task. OSA vs. No OSA accuracies reached 81.28%, 82.05% and 90.26% in the apnea–hypopnea index cutoffs 1, 5 and 10 events/h, respectively. The most relevant information from SpO2 was redundant with ODI 3%, and AF was complementary to them. Thus, the joint analysis of AF and SpO2 enhanced the diagnostic performance of each signal alone using AdaBoost, thereby enabling a potential screening alternative for OSA in children.

Respiratory polygraphy data of children investigated for sleep-disordered breathing with different congenital or respiratory diseases.

This short report describes respiratory indices of polygraphies (PG) performed to investigate several sleep-related disorders of breathing in children. It refers to the work of Michelet et al., Successful home respiratory polygraphy to investigate sleep-disordered breathing in children, Sleep Medicine [1]. Indications for PGs were grouped according to 6 categories: craniofacial malformation, neuromuscular disease, obesity, suspected obstructive sleep apnea (OSA), prematurity, and other. The reported data concern the initial interpretable PGs (N = 289); initial was defined as performed for the first time in any subject. Non-interpretability was defined as absent or unreliable oxygen saturation by pulse oximetry (SpO2), and/or airflow and respiratory inductance plethysmography (RIP) flow trace signals during time analyzed. Analyzed time is reported. In a subset of patients, transcutaneous carbon dioxide partial pressure (ptcCO2) was also measured. Data may be re-used for comparison in future validating research for PGs in children [2].

Detection of Abnormal Respiratory Events with Single Channel ECG and Hybrid Machine Learning Model in Patients with Obstructive Sleep Apnea

Respiratory scoring is an important step in the diagnosis of Obstructive Sleep Apnea (OSA). Airflow, abdolmel-thorax and pulse oximetry signals are obtained with the help of Polysomnography (PSG) device for the respiration scoring stage. These signals are visually scored by a specialist physician. The PSG has several disadvantages: one of them is that a technician is required to use the device. In addition, the records must be taken in the hospital environment. The aim of this study is to develop a new machine learning based hybrid sleep/awake detection method with single channel ECG alternative to respiratory scoring. For this purpose, electrocardiography (ECG) signal of 10 patients with OSA was used. The Heart Rate Variable signal was derived from the ECG signal. Then, QRS components in different frequency bands were obtained from ECG signal by digital filtering. In this way, a total of nine more signals were obtained. Each of the nine signals consists of 25 features, which amounts to a total of 225 features. Fisher feature selection algorithm and Principal Component Analysis (PCA) were used to reduce the number of features. Ultimately the features extracted from the first received signals were classified with Decision Tree, Support Vector Machines, k-Nearest Neighborhood Algorithm and Ensemble classifiers. In addition, the proposed model was checked with the Leave One Out method. At the end of the study, for the detection of apnea, 82.11% accuracy with only 3 features and 85.12% accuracy with 13 features were obtained. The sensitivity and specificity values for the 3 properties are 0.82 and 0.82, respectively. For the 13 properties, 0.85 and 0.86, respectively. These results show that the proposed model can be used for the detection of Respiratory Scoring in the OSA diagnostic process.

Neuromodulation of afferent fibers for Postural Orthostatic Trachycardia Syndrome (POTS)

Postural Orthostatic Trachycardia Syndrome, or POTS, is characterized by low arterial blood pressure, inappropriate cardiovascular and breathing responses to tilt, dizziness and syncope to sudden standing, and a range of other autonomic signs, including thermoregulatory issues in the periphery and frequent manifestations of migraine pain. Interventions are few, and typically include volume repletion by saline and enhanced fluid intake. In the course of evaluating neuromodulatory procedures for migraine pain (n=62), several subjects showed classic POTS symptoms, many of which spontaneously resolved with the intervention for migraine. Non‐invasive neuromodulation of two different areas, one being the auditory canal containing mandibular nerve V3, cranial nerves VII, IX, and X and cervical nerves C2 and C3; and the other being regional stimulation of V1 and V2, together with C2 and C3 sensory areas, caused significant declines in dizziness, hand cooling, and other POTS symptoms. The neuromodulatory device stimulating the auditory canal consisted of bilateral silicon impressions of the auditory canals, each containing a mechanical vibratory device which reached a range of cranial nerves, and especially the IXth (glossopharyngeal) nerve to the carotid baroreceptors. The other neuromodulatory device applied vibrating signals to the cutaneous surface over areas innervated by V1 and V2. The trials consisted of a 10‐min control period, followed by 25‐min stimulation, and finished with a 10‐min control period. Continuous thoracic wall movement, pulse oximetry, and ECG were collected, and beat‐by‐beat blood pressure values were inferred from the pulse transit time of the cardiac R wave and oximetry signal. All procedures were approved by the UCLA Institutional Review Board. Subjects who began the trial with low blood pressure increased systolic and diastolic pressures. In two subjects, repeated trials benefitted critical signs (hand warming, dizziness, syncope to sudden movement). These findings suggest that repeated activation of afferent input to baroreceptors can elicit both short‐ and long‐term effects on blood pressure control. The means by which this “learning" or adaptation of cardiovascular regulatory systems remain unknown, but raise the possibility that a low‐cost, non‐invasive intervention can provide another means to restore effective blood pressure control in a syndrome with few options for treatment approaches.Support or Funding InformationThis research was supported in part by the Fidelity Charitable Nancy Adams and Scott Schoen Fund and the Kraig and Linda Kupiec Family Trust.

Impact of flow disruptions in the delivery room

AimFlow disruptions (FDs) are deviations from the progression of care that compromise safety and efficiency of a specific process. The study aim was to identify the impact of FDs during neonatal resuscitation and determine their association with key process and outcome measures. MethodsProspective observational study of video recorded delivery room resuscitations of neonates <32 weeks gestational age. FDs were classified using an adaptation of Wiegmann’s FD tool. The primary outcome was target oxygenation saturation achievement at 5 min. Secondary outcomes included achieving target saturation at 10 min, time to positive pressure ventilation for initially apnoeic/bradycardic neonates, time to electrocardiogram signal, time to pulse oximetry signal, and time to stable airway. Multivariable logistic regression assessed association between FDs and achieving target saturations adjusting for gestational age and leader. Associations between FDs and time to event outcomes were assessed using Cox proportional hazards models. ResultsBetween 10/2017–7/2018, 32 videos were included. A mean of 52.6 FDs (standard deviation 17.9) occurred per resuscitation. Extraneous FDs were the most common FDs. FDs were associated with an adjusted odds ratio of 0.92 (95% confidence interval [CI] 0.80–1.05) of achieving target saturation at 5 min and 0.94 (95% CI 0.84–1.05) at 10 min. There was no significant evidence to show FDs were associated with time to event outcomes. ConclusionsFDs occurred frequently during neonatal resuscitation. Measuring FDs is a feasible method to assess the impact of human factors in the delivery room and identify modifiable factors and practices to improve patient care.

Towards an anxiety and stress recognition system for academic environments based on physiological features

Background and ObjectiveTraditional methods to determine stress and anxiety in academic environments consist of the application of questionnaires, but the main disadvantage is that the results depend on the students’ self-perception. Being able to detect anxiety-related stress levels in a simple and objective way contributes greatly to dealing with low performance and school drop-out by students. MethodsThe main contribution of this study is to identify the physiological features that could be used as predictors of stressful activities and states of anxiety in academic environments using an Arduino board and low-cost sensors. A test with 21 students was conducted, and a stress-inducing protocol was proposed and 21 physiological features of five signals were analyzed. In addition, the State-Trait Anxiety Inventory (STAI) was used to assess the level of anxiety for each student. Four classifiers were compared to find the physiological feature subset that provides the best accuracy to identify states of stress and anxiety. ResultsThe stress due to activities performed by students can be identified with an accuracy greater than 90% (Kappa = 0.84) using the k-Nearest Neighbors classifier, using data from heart rate, skin temperature and oximetry signals and four physiological features. Meanwhile, the identification of anxiety was achieved with an accuracy greater than 95% (Kappa = 0.90) using the SVM classifier with data from the galvanic skin response (GSR) signal and three physiological features. ConclusionsThe results provide a clue that anxiety detection in academic environments could be done using the analysis of physiological signals instead of STAI test scores. Besides, the results suggest that physiological features could be used to develop stress recognition systems to help teachers to identify the stressful tasks in an academic environment or to develop anxiety recognition systems to help students to control their level of anxiety when they are performing either academic tasks or exams.

Successful home respiratory polygraphy to investigate sleep-disordered breathing in children

ObjectiveSleep-disordered breathing (SDB) in children is common. Interest in sleep tests, such as polygraphy (PG), which can be performed in a non-attended setting, are gaining is increasing. PG has, however, been little studied in children with co-morbidities other than obstructive sleep apnea (OSA), and in particular, if performed in a non-attended setting. We report on the feasibility and interpretability of implementing PGs at home versus in hospital. MethodsPGs were analyzed according to the setting (hospital or home) and sequence (initial or subsequent) in which they were performed. Non-interpretability was defined as absent or unreliable oxygen saturation by pulse oximetry (SpO2), or airflow and respiratory inductance plethysmography flow trace signals during the time analyzed. ResultsWe retrospectively analyzed 400 PGs; 332/400 were initial PGs. Indications were: suspected OSA (65%), obesity (13%), craniofacial malformations (5%), neuromuscular disease (4%), and other (13%) which included prematurity. 16% were recorded in hospitals and 84% at home. The mean age was 5.7 ± 5.8 years and 7.3 ± 4.5 years for the hospital and home groups, respectively. Interpretability was similar in both settings (87%). In the 68 subsequent PGs, interpretability was 84% when performed for follow-up and 96% when repeated for non-interpretability. Non-interpretability was predominantly due to a failure of the SpO2 channel. ConclusionsPG performed at home is both feasible and interpretable for a variety of indications. Non-interpretability was not predictable in association with the setting, anthropometric data, or indication, independently of the sequence (initial or subsequent PG) in which the parameters were analyzed.