Artificial neural network analysis of the oxygen saturation signal enables accurate diagnostics of sleep apnea

Estimating the costs and analysing the precision of several diagnostic and treatment approaches for obstructive sleep apnea patients in the Netherlands, using timed automata modelling

Obstructive Sleep Apnea (OSA) is a (highly) prevalent medical condition, linked to severe negative health consequences. In the Netherlands, diagnosis of OSA presently has long waiting times and both diagnosis and treatment have high costs. This article introduces a so-called timed automata model (UPPAAL tool) for analysing diagnostic and treatment approaches for OSA. This model is used for assessing the Dutch current traditional approach, as well as multiple alternative approaches for OSA diagnosis and treatment. The analysis shows that one alternative approach can lower the costs and waiting lists, while maintaining diagnostic precision. In this manuscript the best alternative approach is a combination of Oxygen Desaturation Index (ODI) measurement obtained by nocturnal pulse oximetry and a questionnaire to diagnose the patient. Of course, it is important that, although timed automata modelling is a reliable tool, these outcomes are meant to start a discussion regarding the above-mentioned problems and are not proven outcomes yet. The healthcare system in the Netherlands is in danger of becoming unaffordable. Therefore, this initial exploration has been carried out to see whether alternatives to the current OSA-care can be devised.

Advancements in Obstructive Sleep Apnea Diagnosis and Screening Through Artificial Intelligence: A Systematic Review

BACKGROUND

Obstructive sleep apnea (OSA) is a prevalent yet underdiagnosed condition associated with a major healthcare burden. Current diagnostic tools, such as full-night polysomnography (PSG), pose a limited accessibility to diagnosis due to their elevated costs. Recent advances in Artificial Intelligence (AI), including Machine Learning (ML) and deep learning (DL) algorithms, offer novel potential tools for an accurate OSA screening and diagnosis. This systematic review evaluates articles employing AI-powered models for OSA screening and diagnosis in the last decade.

METHODS

A comprehensive electronic search was performed on PubMed/MEDLINE, Google Scholar, and SCOPUS databases. The included studies were original articles written in English, reporting the use of ML algorithms to diagnose and predict OSA in suspected patients. The last search was performed in June 2024. This systematic review is registered in PROSPERO (Registration ID: CRD42024563059).

RESULTS

Sixty-five articles, involving data from 109,046 patients, met the inclusion criteria. Due to the heterogeneity of the algorithms, outcomes were analyzed into six sections (anthropometric indexes, imaging, electrocardiographic signals, respiratory signals, and oximetry and miscellaneous signals). AI algorithms demonstrated significant improvements in OSA detection, with accuracy, sensitivity, and specificity often exceeding traditional tools. In particular, anthropometric indexes were most widely used, especially in logistic regression-powered algorithms.

CONCLUSIONS

The application of AI algorithms to OSA diagnosis and screening has great potential to improve patient outcomes, increase early detection, and lessen the load on healthcare systems. However, rigorous validation and standardization efforts must be made to standardize datasets.

Thermal performance prediction of a V-trough solar water heater with a modified twisted tape using ANFIS, G.L.R., R.T. and SVM models of machine learning

Four distinct neural models were used to evaluate the efficiency of a V-trough solar water heater (VTSWH) equipped with square-cut twisted tape (SCTT) and V-cut twisted tape (VCTT) at two different twist ratios, 3 and 5. The objective of this study was the use of ANFIS (Adaptive Neuro-Fuzzy Inference System), G.L.R. (Generalised linear regression), R.T. (Regression tree), and SVM (Support Vector Machine). A total of 162 data sets were acquired for these models through a variety of trials. Outdoor experiments were done using a twist ratio of Y = 3 and Y = 5, using both SCTT and VCTT. The models included eight distinct variables: ambient temperature, water mass flow rate, water intake temperature, water exit temperature, absorber plate temperature, tube temperature, solar intensity, and twist ratio. The dependent variables in this study are the Nusselt number (Nu), friction factor (FF), and efficiency (η). 130 datasets were chosen for training purposes, while 32 were used for testing. Using the ANFIS, G.L.R., R.T., and SVM techniques, the correlation coefficient (R2) values for Nusselt number were 0.9990, 0.9961, 0.9562, and 0.9280 for friction factor 0.9966, 0.9683, 0.9810, and 0.9560, and for efficiency 0.9997, 0.9976, 0.9845, and 0.9614, respectively. Comparing all models shows that ANFIS is the most effective of the four strategies studied. The ANFIS model outperformed the other models regarding Nu, FF, and η, with RMSE values of 0.0805, 0.0.0004, and 0.4534. According to the above data, the VTSWH thermal performance predicted using the ANFIS approach has the highest accuracy.

Obstructive sleep apnea diagnosis and beyond using portable monitors

Obstructive sleep apnea (OSA) is a chronic sleep and breathing disorder with significant health complications, including cardiovascular disease and neurocognitive impairments. To ensure timely treatment, there is a need for a portable, accurate and rapid method of diagnosing OSA. This review examines the use of various physiological signals used in the detection of respiratory events and evaluates their effectiveness in portable monitors (PM) relative to gold standard polysomnography. The primary objective is to explore the relationship between these physiological parameters and OSA, their application in calculating the apnea hypopnea index (AHI), the standard metric for OSA diagnosis, and the derivation of non-AHI metrics that offer additional diagnostic value. It is found that increasing the number of parameters in PMs does not necessarily improve OSA detection. Several factors can cause performance variations among different PMs, even if they extract similar signals. The review also highlights the potential of PMs to be used beyond OSA diagnosis. These devices possess parameters that can be utilized to obtain endotypic and other non-AHI metrics, enabling improved characterization of the disorder and personalized treatment strategies. Advancements in PM technology, coupled with thorough evaluation and validation of these devices, have the potential to revolutionize OSA diagnosis, personalized treatment, and ultimately improve health outcomes for patients with OSA. By identifying the key factors influencing performance and exploring the application of PMs beyond OSA diagnosis, this review aims to contribute to the ongoing development and utilization of portable, efficient, and effective diagnostic tools for OSA.

Application of artificial intelligence in the diagnosis of sleep apnea

STUDY OBJECTIVES

Machine learning (ML) models have been employed in the setting of sleep disorders. This review aims to summarize the existing data about the role of ML techniques in the diagnosis, classification, and treatment of sleep-related breathing disorders.

METHODS

A systematic search in Medline, EMBASE, and Cochrane databases through January 2022 was performed.

RESULTS

Our search strategy revealed 132 studies that were included in the systematic review. Existing data show that ML models have been successfully used for diagnostic purposes. Specifically, ML models showed good performance in diagnosing sleep apnea using easily obtained features from the electrocardiogram, pulse oximetry, and sound signals. Similarly, ML showed good performance for the classification of sleep apnea into obstructive and central categories, as well as predicting apnea severity. Existing data show promising results for the ML-based guided treatment of sleep apnea. Specifically, the prediction of outcomes following surgical treatment and optimization of continuous positive airway pressure therapy can be guided by ML models.

CONCLUSIONS

The adoption and implementation of ML in the field of sleep-related breathing disorders is promising. Advancements in wearable sensor technology and ML models can help clinicians predict, diagnose, and classify sleep apnea more accurately and efficiently.

CITATION

Bazoukis G, Bollepalli SC, Chung CT, et al. Application of artificial intelligence in the diagnosis of sleep apnea. J Clin Sleep Med. 2023;19(7):1337-1363.

Utilizing Envelope Analysis of a Nasal Pressure Signal for Sleep Apnea Severity Estimation

Obstructive sleep apnea (OSA) severity assessment is based on manually scored respiratory events and their arbitrary definitions. Thus, we present an alternative method to objectively evaluate OSA severity independently of the manual scorings and scoring rules. A retrospective envelope analysis was conducted on 847 suspected OSA patients. Four parameters were calculated from the difference between the nasal pressure signal's upper and lower envelopes: average (AV), median (MD), standard deviation (SD), and coefficient of variation (CoV). We computed the parameters from the entirety of the recorded signals to perform binary classifications of patients using three different apnea-hypopnea index (AHI) thresholds (5-15-30). Additionally, the calculations were undertaken in 30-second epochs to estimate the ability of the parameters to detect manually scored respiratory events. Classification performances were assessed with areas under the curves (AUCs). As a result, the SD (AUCs ≥ 0.86) and CoV (AUCs ≥ 0.82) were the best classifiers for all AHI thresholds. Furthermore, non-OSA and severe OSA patients were separated well with SD (AUC = 0.97) and CoV (AUC = 0.95). Respiratory events within the epochs were identified moderately with MD (AUC = 0.76) and CoV (AUC = 0.82). In conclusion, envelope analysis is a promising alternative method by which to assess OSA severity without relying on manual scoring or the scoring rules of respiratory events.

Research Progress of Respiratory Disease and Idiopathic Pulmonary Fibrosis Based on Artificial Intelligence

Machine Learning (ML) is an algorithm based on big data, which learns patterns from the previously observed data through classifying, predicting, and optimizing to accomplish specific tasks. In recent years, there has been rapid development in the field of ML in medicine, including lung imaging analysis, intensive medical monitoring, mechanical ventilation, and there is need for intubation etiology prediction evaluation, pulmonary function evaluation and prediction, obstructive sleep apnea, such as biological information monitoring and so on. ML can have good performance and is a great potential tool, especially in the imaging diagnosis of interstitial lung disease. Idiopathic pulmonary fibrosis (IPF) is a major problem in the treatment of respiratory diseases, due to the abnormal proliferation of fibroblasts, leading to lung tissue destruction. The diagnosis mainly depends on the early detection of imaging and early treatment, which can effectively prolong the life of patients. If the computer can be used to assist the examination results related to the effects of fibrosis, a timely diagnosis of such diseases will be of great value to both doctors and patients. We also previously proposed a machine learning algorithm model that can play a good clinical guiding role in early imaging prediction of idiopathic pulmonary fibrosis. At present, AI and machine learning have great potential and ability to transform many aspects of respiratory medicine and are the focus and hotspot of research. AI needs to become an invisible, seamless, and impartial auxiliary tool to help patients and doctors make better decisions in an efficient, effective, and acceptable way. The purpose of this paper is to review the current application of machine learning in various aspects of respiratory diseases, with the hope to provide some help and guidance for clinicians when applying algorithm models.

Desaturation event scoring criteria affect the perceived severity of nocturnal hypoxic load

OBJECTIVES/BACKGROUND

Interest in using blood oxygen desaturations in the diagnostics of sleep apnea has risen in recent years. However, no standardized criteria for desaturation scoring exist which complicates the drawing of solid conclusions from literature.

PATIENTS/METHODS

We investigated how different desaturation scoring criteria affect the severity of nocturnal hypoxic load and the prediction of impaired daytime vigilance in 845 patients. Desaturations were scored based on three features: 1) minimum oxygen saturation drop during the event (2-20%, 1% interval), 2) minimum duration of the event (2-20s, 1s interval), and 3) maximum plateau duration within the event (5-60s, 5s interval), resulting in 4332 different scoring criteria. The hypoxic load was described with oxygen desaturation index (ODI), desaturation severity (DesSev), and desaturation duration (DesDur) parameters. Association between hypoxic load and impaired vigilance was investigated with covariate-adjusted area under curve (AUC) analyses by dividing patients into normal (≤5 lapses) and impaired (≥36 lapses) vigilance groups based on psychomotor vigilance task performance.

RESULTS

The severity of hypoxic load varied greatly between different scoring criteria. For example, median ODI ranged between 0.4 and 12.9 events/h, DesSev 0.01-0.23 %-point, and DesDur 0.3-9.6 %-point when the minimum transient drop criterion of 3% was used and other two features were altered. Overall, the minimum transient drop criterion had the largest effect on parameter values. All models with differently determined parameters predicted impaired vigilance moderately (AUC = 0.722-0.734).

CONCLUSIONS

Desaturation scoring criteria greatly affected the severity of hypoxic load. However, the difference in the prediction of impaired vigilance between different criteria was rather small.

Association between obstructive sleep apnea (OSA) and atrial fibrillation (AF): A dose-response meta-analysis

BACKGROUND

Refractory hypoxemia episodes are characteristic of obstructive sleep apnea (OSA). Patients with OSA suffer from oxidative stress in all systems. Atrial fibrillation (AF) is a type of arrhythmia that may be induced by OSA. In this study, we explored the dose-response relationship between OSA and AF. Our research provides the basis for a novel approach to AF prevention.

METHODS

We screened four databases (PubMed, Embase, the Cochrane Library, and Web of Science) for observational studies on OSA and AF. Studies were collected from database establishment to November 2020. We performed a traditional subgroup meta-analysis. Linear and spline dose-response models were applied to assess the association between the apnea-hypopnea index, an indicator of OSA severity, and the risk of AF. Review Manager version 5.3 software and Stata 16.0 were used for the analysis.

RESULTS

Sixteen observational studies were included in the study. We excluded a study from the conventional meta-analysis. In the subgroup analysis, the odds ratios for new onset AF for no obvious reason, new onset AF after surgical operations, such as coronary artery bypass grafting, and AF after ablation treatment were 1.71 (95% CI 1.37-2.13, P < .05), 2.65 (95% CI 2.32-3.01, P < .05), and 2.93 (95% CI 2.47-3.49, P < .05), respectively. Linear dose-response meta-analysis results revealed that the risk of AF increased with increasing apnea-hypopnea index value.

CONCLUSION

Through dose-response meta-analysis, we found a potential dose-response relationship between OSA severity and the risk of AF. This relationship should be considered in interventions aimed at AF prevention in the future.

The Sleep Revolution project: the concept and objectives

Obstructive sleep apnea is linked to severe health consequences such as hypertension, daytime sleepiness, and cardiovascular disease. Nearly a billion people are estimated to have obstructive sleep apnea with a substantial economic burden. However, the current diagnostic parameter of obstructive sleep apnea, the apnea-hypopnea index, correlates poorly with related comorbidities and symptoms. Obstructive sleep apnea severity is measured by counting respiratory events, while other physiologically relevant consequences are ignored. Furthermore, as the clinical methods for analysing polysomnographic signals are outdated, laborious, and expensive, most patients with obstructive sleep apnea remain undiagnosed. Therefore, more personalised diagnostic approaches are urgently needed. The Sleep Revolution, funded by the European Union's Horizon 2020 Research and Innovation Programme, aims to tackle these shortcomings by developing machine learning tools to better estimate obstructive sleep apnea severity and phenotypes. This allows for improved personalised treatment options, including increased patient participation. Also, implementing these tools will alleviate the costs and increase the availability of sleep studies by decreasing manual scoring labour. Finally, the project aims to design a digital platform that functions as a bridge between researchers, patients, and clinicians, with an electronic sleep diary, objective cognitive tests, and questionnaires in a mobile application. These ambitious goals will be achieved through extensive collaboration between 39 centres, including expertise from sleep medicine, computer science, and industry and by utilising tens of thousands of retrospectively and prospectively collected sleep recordings. With the commitment of the European Sleep Research Society and Assembly of National Sleep Societies, the Sleep Revolution has the unique possibility to create new standardised guidelines for sleep medicine.

Barriers of artificial intelligence implementation in the diagnosis of obstructive sleep apnea

Background

Obstructive sleep apnea is a common clinical condition and has a significant impact on the health of patients if untreated. The current diagnostic gold standard for obstructive sleep apnea is polysomnography, which is labor intensive, requires specialists to utilize, expensive, and has accessibility challenges. There are also challenges with awareness and identification of obstructive sleep apnea in the primary care setting. Artificial intelligence systems offer the opportunity for a new diagnostic approach that addresses the limitations of polysomnography and ultimately benefits patients by streamlining the diagnostic expedition.

Main body

The purpose of this project is to elucidate the barriers that exist in the implementation of artificial intelligence systems into the diagnostic context of obstructive sleep apnea. It is essential to understand these challenges in order to proactively create solutions and establish an efficient adoption of this new technology. The literature regarding the evolution of the diagnosis of obstructive sleep apnea, the role of artificial intelligence in the diagnosis, and the barriers in artificial intelligence implementation was reviewed and analyzed.

Conclusion

The barriers identified were categorized into different themes including technology, data, regulation, human resources, education, and culture. Many of these challenges are ubiquitous across artificial intelligence implementation in any medical diagnostic setting. Future research directions include developing solutions to the barriers presented in this project.

Graphical abstract

Pulse Oximetry: The Working Principle, Signal Formation, and Applications

Pulse oximeters are routinely used in various medical-grade and consumer-grade applications. They can be used to estimate, for example, blood oxygen saturation, autonomic nervous system activity and cardiac function, blood pressure, sleep quality, and recovery through the recording of photoplethysmography signal. Medical-grade devices often record red and infra-red light-based photoplethysmography signals while smartwatches and other consumer-grade devices usually rely on a green light. At its simplest, a pulse oximeter can consist of one or two photodiodes and a photodetector attached, for example, a fingertip or earlobe. These sensors are used to record light absorption in a medium as a function of time. This time-varying absorption information is used to form a photoplethysmography signal. In this chapter, we discuss the working principles of pulse oximeters and the formation of the photoplethysmography signal. We will further discuss the advantages and disadvantages of pulse oximeters, which kind of applications exist in the medical field, and how pulse oximeters are utilized in daily health monitoring.

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.

Improving Machine Learning Technology in the Field of Sleep

The authors discuss the challenges of machine- and deep learning-based automatic analysis of obstructive sleep apnea with respect to known issues with the signal interpretation, patient physiology, and the apnea-hypopnea index. Their goal is to provide guidance for sleep and machine learning professionals working in this area of sleep medicine. They suggest that machine learning approaches may well be better targeted at examining and attempting to improve the diagnostic criteria, in order to build a more nuanced understanding of the detailed circumstances surrounding OSA, rather than merely attempting to reproduce human scoring.

Self-Applied Home Sleep Recordings: The Future of Sleep Medicine

Sleep disorders form a massive global health burden and there is an increasing need for simple and cost-efficient sleep recording devices. Recent machine learning-based approaches have already achieved scoring accuracy of sleep recordings on par with manual scoring, even with reduced recording montages. Simple and inexpensive monitoring over multiple consecutive nights with automatic analysis could be the answer to overcome the substantial economic burden caused by poor sleep and enable more efficient initial diagnosis, treatment planning, and follow-up monitoring for individuals suffering from sleep disorders.

PSG-Audio, a scored polysomnography dataset with simultaneous audio recordings for sleep apnea studies

The sleep apnea syndrome is a chronic condition that affects the quality of life and increases the risk of severe health conditions such as cardiovascular diseases. However, the prevalence of the syndrome in the general population is considered to be heavily underestimated due to the restricted number of people seeking diagnosis, with the leading cause for this being the inconvenience of the current reference standard for apnea diagnosis: Polysomnography. To enhance patients' awareness of the syndrome, a great endeavour is conducted in the literature. Various home-based apnea detection systems are being developed, profiting from information in a restricted set of polysomnography signals. In particular, breathing sound has been proven highly effective in detecting apneic events during sleep. The development of accurate systems requires multitudinous datasets of audio recordings and polysomnograms. In this work, we provide the first open access dataset, comprising 212 polysomnograms along with synchronized high-quality tracheal and ambient microphone recordings. We envision this dataset to be widely used for the development of home-based apnea detection techniques and frameworks.

Automatic Respiratory Event Scoring in Obstructive Sleep Apnea Using a Long Short-Term Memory Neural Network

The diagnosis of obstructive sleep apnea is based on daytime symptoms and the frequency of respiratory events during the night. The respiratory events are scored manually from polysomnographic recordings, which is time-consuming and expensive. Therefore, automatic scoring methods could considerably improve the efficiency of sleep apnea diagnostics and release the resources currently needed for manual scoring to other areas of sleep medicine. In this study, we trained a long short-term memory neural network for automatic scoring of respiratory events using input signals from peripheral blood oxygen saturation, thermistor-airflow, nasal pressure -airflow, and thorax respiratory effort. The signals were extracted from 887 in-lab polysomnography recordings. 787 patients with suspected sleep apnea were used to train the neural network and 100 patients were used as an independent test set. The epoch-wise agreement between manual and automatic neural network scoring was high (88.9%, κ = 0.728). In addition, the apnea-hypopnea index (AHI) calculated from the automated scoring was close to the manually determined AHI with a mean absolute error of 3.0 events/hour and an intraclass correlation coefficient of 0.985. The neural network approach for automatic scoring of respiratory events achieved high accuracy and good agreement with manual scoring. The presented neural network could be used for analysis of large research datasets that are unfeasible to score manually, and has potential for clinical use in the future In addition, since the neural network scores individual respiratory events, the automatic scoring can be easily reviewed manually if desired.

Accelerometry-derived respiratory index estimating apnea-hypopnea index for sleep apnea screening

BACKGROUND AND OBJECTIVE

Sleep Apnea Syndrome (SAS) is a multimorbid chronic disease with individual and societal deleterious consequences. Polysomnography (PSG) is the multi-parametric reference diagnostic tool that allows a manual quantification of the apnea-hypopnea index (AHI) to assess SAS severity. The burden of SAS is affecting nearly one billion people worldwide explaining that SAS remains largely under-diagnosed and undertreated. The development of an easy to use and automatic solution for early detection and screening of SAS is highly desirable.

METHODS

We proposed an Accelerometry-Derived Respiratory index (ADR) solution based on a dual accelerometry system for airflow estimation included in a machine learning process. It calculated the AHI thanks to a RUSBoosted Tree model and used physiological and explanatory specifically developed features. The performances of this method were evaluated against a configuration using gold-standard PSG signals on a database of 28 subjects.

RESULTS

The AHI estimation accuracy, specificity and sensitivity of the ADR index were 89%, 100% and 80% respectively. The added value of the specifically developed features was also demonstrated.

CONCLUSION

Overnight physiological monitoring with the proposed ADR solution using a machine learning approach provided a clinically relevant estimate of AHI for SAS screening. The physiological component of the solution has a real interest for improving performance and facilitating physician's adhesion to an automatic AHI estimation.

A Convolutional Neural Network Architecture to Enhance Oximetry Ability to Diagnose Pediatric Obstructive Sleep Apnea

This study aims at assessing the usefulness of deep learning to enhance the diagnostic ability of oximetry in the context of automated detection of pediatric obstructive sleep apnea (OSA). A total of 3196 blood oxygen saturation (SpO2) signals from children were used for this purpose. A convolutional neural network (CNN) architecture was trained using 20-min SpO2 segments from the training set (859 subjects) to estimate the number of apneic events. CNN hyperparameters were tuned using Bayesian optimization in the validation set (1402 subjects). This model was applied to three test sets composed of 312, 392, and 231 subjects from three independent databases, in which the apnea-hypopnea index (AHI) estimated for each subject (AHICNN) was obtained by aggregating the output of the CNN for each 20-min SpO2 segment. AHICNN outperformed the 3% oxygen desaturation index (ODI3), a clinical approach, as well as the AHI estimated by a conventional feature-engineering approach based on multi-layer perceptron (AHIMLP). Specifically, AHICNN reached higher four-class Cohen's kappa in the three test databases than ODI3 (0.515 vs 0.417, 0.422 vs 0.372, and 0.423 vs 0.369) and AHIMLP (0.515 vs 0.377, 0.422 vs 0.381, and 0.423 vs 0.306). In addition, our proposal outperformed state-of-the-art studies, particularly for the AHI severity cutoffs of 5 e/h and 10 e/h. This suggests that the information automatically learned from the SpO2 signal by deep-learning techniques helps to enhance the diagnostic ability of oximetry in the context of pediatric OSA.

Estimation of Apnea-Hypopnea Index using Deep Learning on 3D Craniofacial Scans

Obstructive sleep apnea (OSA) is characterized by decreased breathing events that occur through the night, with severity reported as the apnea-hypopnea index (AHI), which is associated with certain craniofacial features. In this study, we used data from 1366 patients collected as part of Stanford Technology Analytics and Genomics in Sleep (STAGES) across 11 US and Canadian sleep clinics and analyzed 3D craniofacial scans with the goal of predicting AHI, as measured using gold standard nocturnal polysomnography (PSG). First, the algorithm detects pre-specified landmarks on mesh objects and aligns scans in 3D space. Subsequently, 2D images and depth maps are generated by rendering and rotating scans by 45-degree increments. Resulting images were stacked as channels and used as input to multi-view convolutional neural networks, which were trained and validated in a supervised manner to predict AHI values derived from PSGs. The proposed model achieved a mean absolute error of 11.38 events/hour, a Pearson correlation coefficient of 0.4, and accuracy for predicting OSA of 67% using 10-fold cross-validation. The model improved further by adding patient demographics and variables from questionnaires. We also show that the model performed at the level of three sleep medicine specialists, who used clinical experience to predict AHI based on 3D scan displays. Finally, we created topographic displays of the most important facial features used by the model to predict AHI, showing importance of the neck and chin area. The proposed algorithm has potential to serve as an inexpensive and efficient screening tool for individuals with suspected OSA.

Longer and Deeper Desaturations Are Associated With the Worsening of Mild Sleep Apnea: The Sleep Heart Health Study

Study Objectives

Obesity, older age, and male sex are recognized risk factors for sleep apnea. However, it is unclear whether the severity of hypoxic burden, an essential feature of sleep apnea, is associated with the risk of sleep apnea worsening. Thus, we investigated our hypothesis that the worsening of sleep apnea is expedited in individuals with more severe desaturations.

Methods

The blood oxygen saturation (SpO₂) signals of 805 Sleep Heart Health Study participants with mild sleep apnea [5 ≤ oxygen desaturation index (ODI) < 15] were analyzed at baseline and after a mean follow-up time of 5.2 years. Linear regression analysis, adjusted for relevant covariates, was utilized to study the association between baseline SpO₂-derived parameters and change in sleep apnea severity, determined by a change in ODI. SpO₂-derived parameters, consisting of ODI, desaturation severity (DesSev), desaturation duration (DesDur), average desaturation area (avg. DesArea), and average desaturation duration (avg. DesDur), were standardized to enable comparisons between the parameters.

Results

In the group consisting of both men and women, avg. DesDur (β = 1.594, p = 0.001), avg. DesArea (β = 1.316, p = 0.004), DesDur (β = 0.998, p = 0.028), and DesSev (β = 0.928, p = 0.040) were significantly associated with sleep apnea worsening, whereas ODI was not (β = -0.029, p = 0.950). In sex-stratified analysis, avg. DesDur (β = 1.987, p = 0.003), avg. DesArea (β = 1.502, p = 0.024), and DesDur (β = 1.374, p = 0.033) were significantly associated with sleep apnea worsening in men.

Conclusion

Longer and deeper desaturations are more likely to expose a patient to the worsening of sleep apnea. This information could be useful in the planning of follow-up monitoring or lifestyle counseling in the early stage of the disease.

Deep Learning for Diagnosis and Classification of Obstructive Sleep Apnea: A Nasal Airflow-Based Multi-Resolution Residual Network

PURPOSE

This study evaluated a novel approach for diagnosis and classification of obstructive sleep apnea (OSA), called Obstructive Sleep Apnea Smart System (OSASS), using residual networks and single-channel nasal pressure airflow signals.

METHODS

Data were collected from the sleep center of the First Affiliated Hospital, Sun Yat-sen University, and the Integrative Department of Guangdong Province Traditional Chinese Medical Hospital. We developed a new model called the multi-resolution residual network (Mr-ResNet) based on a residual network to detect nasal pressure airflow signals recorded by polysomnography (PSG) automatically. The performance of the model was assessed by its sensitivity, specificity, accuracy, and F1-score. We built OSASS based on Mr-ResNet to estimate the apnea‒hypopnea index (AHI) and to classify the severity of OSA, and compared the agreement between OSASS output and the registered polysomnographic technologist (RPSGT) score, assessed by two technologists.

RESULTS

In the primary test set, the sensitivity, specificity, accuracy, and F1-score of Mr-ResNet were 90.8%, 90.5%, 91.2%, and 90.5%, respectively. In the independent test set, the Spearman correlation for AHI between OSASS and the RPSGT score determined by two technologists was 0.94 (p < 0.001) and 0.96 (p < 0.001), respectively. Cohen's Kappa scores for classification between OSASS and the two technologists' scores were 0.81 and 0.84, respectively.

CONCLUSION

Our results indicated that OSASS can automatically diagnose and classify OSA using signals from a single-channel nasal pressure airflow, which is consistent with polysomnographic technologists' findings. Thus, OSASS holds promise for clinical application.

Neural network analysis of nocturnal SpO2 signal enables easy screening of sleep apnea in patients with acute cerebrovascular disease

Current diagnostics of sleep apnea relies on the time-consuming manual analysis of complex sleep registrations, which is impractical for routine screening in hospitalized patients with a high probability for sleep apnea, e.g. those experiencing acute stroke or transient ischemic attacks (TIA). To overcome this shortcoming, we aimed to develop a convolutional neural network (CNN) capable of estimating the severity of sleep apnea in acute stroke and TIA patients based solely on the nocturnal oxygen saturation (SpO₂) signal. The CNN was trained with SpO₂ signals derived from 1379 home sleep apnea tests (HSAT) of suspected sleep apnea patients and tested with SpO₂ signals of 77 acute ischemic stroke or TIA patients. The CNN's performance was tested by comparing the estimated respiratory event index (REI) and oxygen desaturation index (ODI) with manually obtained values. Median estimation errors for REI and ODI in patients with stroke or TIA were 1.45 events/hour and 0.61 events/hour, respectively. Furthermore, based on estimated REI and ODI, 77.9% and 88.3% of these patients were classified into the correct sleep apnea severity categories. The sensitivity and specificity to identify sleep apnea (REI > 5 events/hour) were 91.8% and 78.6%, respectively. Moderate-to-severe sleep apnea was detected (REI > 15 events/hour) with sensitivity of 92.3% and specificity of 96.1%. The CNN analysis of the SpO₂ signal has great potential as a simple screening tool for sleep apnea. This novel automatic method accurately detects sleep apnea in acute cerebrovascular disease patients and facilitates their referral for a differential diagnostic HSAT or polysomnography evaluation.

Estimating daytime sleepiness with previous night electroencephalography, electrooculography, and electromyography spectrograms in patients with suspected sleep apnea using a convolutional neural network

A common symptom of obstructive sleep apnea (OSA) is excessive daytime sleepiness (EDS). The gold standard test for EDS is the multiple sleep latency test (MSLT). However, due to its high cost, MSLT is not routinely conducted for OSA patients and EDS is instead evaluated using sleep questionnaires. This is problematic however, since sleep questionnaires are subjective and correlate poorly with the MSLT. Therefore, new objective tools are needed for reliable evaluation of EDS. The aim of this study was to test our hypothesis that EDS can be estimated with neural network analysis of previous night polysomnographic signals. We trained a convolutional neural network (CNN) classifier using electroencephalography, electrooculography, and chin electromyography signals from 2,014 patients with suspected OSA. The CNN was trained to classify the patients into four sleepiness categories based on their mean sleep latency (MSL); severe (MSL < 5min), moderate (5 ≤ MSL < 10), mild (10 ≤ MSL < 15), and normal (MSL ≥ 15). The CNN classified patients to the four sleepiness categories with an overall accuracy of 60.6% and Cohen's kappa value of 0.464. In two-group classification scheme with sleepy (MSL < 10 min) and non-sleepy (MSL ≥ 10) patients, the CNN achieved an accuracy of 77.2%, with sensitivity of 76.5%, and specificity of 77.9%. Our results show that previous night's polysomnographic signals can be used for objective estimation of EDS with at least moderate accuracy. Since the diagnosis of OSA is currently confirmed by polysomnography, the classifier could be used simultaneously to get an objective estimate of the daytime sleepiness with minimal extra workload.

Increased nocturnal arterial pulsation frequencies of obstructive sleep apnoea patients is associated with an increased number of lapses in a psychomotor vigilance task

Objectives

Besides hypoxaemia severity, heart rate variability has been linked to cognitive decline in obstructive sleep apnoea (OSA) patients. Thus, our aim was to examine whether the frequency domain features of a nocturnal photoplethysmogram (PPG) can be linked to poor performance in the psychomotor vigilance task (PVT).

Methods

PPG signals from 567 suspected OSA patients, extracted from Type 1 diagnostic polysomnography, and corresponding results of PVT were retrospectively examined. The frequency content of complete PPGs was determined, and analyses were conducted separately for men (n=327) and women (n=240). Patients were grouped into PVT performance quartiles based on the number of lapses (reaction times ≥500 ms) and within-test variation in reaction times. The best-performing (Q1) and worst-performing (Q4) quartiles were compared due the lack of clinical thresholds in PVT.

Results

We found that the increase in arterial pulsation frequency (APF) in both men and women was associated with a higher number of lapses. Higher APF was also associated with higher within-test variation in men, but not in women. Median APF (β=0.27, p=0.01), time spent under 90% saturation (β=0.05, p<0.01), female sex (β=1.29, p<0.01), older age (β=0.03, p<0.01) and subjective sleepiness (β=0.07, p<0.01) were significant predictors of belonging to Q4 based on lapses. Only female sex (β=0.75, p<0.01) and depression (β=0.91, p<0.02) were significant predictors of belonging to Q4 based on the within-test variation.

Conclusions

In conclusion, increased APF in PPG provides a possible polysomnography indicator for deteriorated vigilance especially in male OSA patients. This finding highlights the connection between cardiorespiratory regulation, vigilance and OSA. However, our results indicate substantial sex-dependent differences that warrant further prospective studies.

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.

Applications of artificial intelligence and machine learning in respiratory medicine

The past 5 years have seen an explosion of interest in the use of artificial intelligence (AI) and machine learning techniques in medicine. This has been driven by the development of deep neural networks (DNNs)-complex networks residing in silico but loosely modelled on the human brain-that can process complex input data such as a chest radiograph image and output a classification such as 'normal' or 'abnormal'. DNNs are 'trained' using large banks of images or other input data that have been assigned the correct labels. DNNs have shown the potential to equal or even surpass the accuracy of human experts in pattern recognition tasks such as interpreting medical images or biosignals. Within respiratory medicine, the main applications of AI and machine learning thus far have been the interpretation of thoracic imaging, lung pathology slides and physiological data such as pulmonary function tests. This article surveys progress in this area over the past 5 years, as well as highlighting the current limitations of AI and machine learning and the potential for future developments.