Machine learning for image-based detection of patients with obstructive sleep apnea: an exploratory study

Abstract Purpose In 2-dimensional lateral cephalometric radiographs, patients with severe obstructive sleep apnea (OSA) exhibit a more crowded oropharynx in comparison with non-OSA. We tested the hypothesis that machine learning, an application of artificial intelligence (AI), could be used to detect patients with severe OSA based on 2-dimensional images. Methods A deep convolutional neural network was developed (n = 1258; 90%) and tested (n = 131; 10%) using data from 1389 (100%) lateral cephalometric radiographs obtained from individuals diagnosed with severe OSA (n = 867; apnea hypopnea index > 30 events/h sleep) or non-OSA (n = 522; apnea hypopnea index < 5 events/h sleep) at a single center for sleep disorders. Three kinds of data sets were prepared by changing the area of interest using a single image: the original image without any modification (full image), an image containing a facial profile, upper airway, and craniofacial soft/hard tissues (main region), and an image containing part of the occipital region (head only). A radiologist also performed a conventional manual cephalometric analysis of the full image for comparison. Results The sensitivity/specificity was 0.87/0.82 for full image, 0.88/0.75 for main region, 0.71/0.63 for head only, and 0.54/0.80 for the manual analysis. The area under the receiver-operating characteristic curve was the highest for main region 0.92, for full image 0.89, for head only 0.70, and for manual cephalometric analysis 0.75. Conclusions A deep convolutional neural network identified individuals with severe OSA with high accuracy. Future research on this concept using AI and images can be further encouraged when discussing triage of OSA.

Download Full-text

0594 Can a Deep Convolutional Neural Network Extract Diagnostic Information on Obstructive Sleep Apnea from Images?

SLEEP ◽

10.1093/sleep/zsaa056.591 ◽

2020 ◽

Vol 43 (Supplement_1) ◽

pp. A227-A227

Author(s):

S Tsuiki ◽

T Nagaoka ◽

T Fukuda ◽

Y Sakamoto ◽

F R Almeida ◽

...

Keyword(s):

Neural Network ◽

Obstructive Sleep Apnea ◽

Sleep Apnea ◽

Convolutional Neural Network ◽

Diagnostic Information ◽

Deep Convolutional Neural Network ◽

Apnea Hypopnea Index ◽

Manual Analysis ◽

Obstructive Sleep ◽

Main Region

Abstract Introduction Lateral cephalometric radiography is a simple way to provide craniofacial soft/hard tissue profiles specific for patients with obstructive sleep apnea (OSA) and may thus offer diagnostic information on the disease. We hypothesized that a machine learning technology, a deep convolutional neural network (DCNN), could make it possible to detect OSA based solely on lateral cephalometric radiographs without the need for either large amounts of subjective/laboratory data or skilled analyses. Methods In this diagnostic study, a DCNN was developed (n=1,258) and tested (n=131) using data from 1,389 lateral cephalometric radiographs obtained from individuals diagnosed with severe OSA (n=867; apnea hypopnea index >30/hour) or non-OSA (n=522; apnea hypopnea index < 5) at a single center for sleep disorders from March, 2006 to February, 2017. Three kinds of data sets were prepared by changing the area of interest using a single image; original image without any modification (Full Image), image containing a facial profile, upper airway, craniofacial soft/hard tissues, and image containing part of the occipital region (upper left corner of the image; Head Only). A radiologist and an orthodontist also performed a manual cephalometric analysis of the Full Image for comparison. Observers were blinded to the patient groupings. Data analysis was performed from April, 2018 to August, 2019. When the predictive score obtained from the DCNN analysis exceeded the threshold (0.50), the patient was judged to have OSA. The primary outcome was diagnostic accuracy in terms of area under the receiver-operating characteristic curve. Results The sensitivity/specificity was 0.87/0.82 for Full Image, 0.88/0.75 for Main Region, 0.71/0.63 for Head Only, and 0.54/0.80 for the manual analysis. The area under the curve was the highest for Main Region (0.92): 0.89 for Full Image, 0.70 for Head Only, and 0.75 for the manual analysis. Conclusion A DCNN identified individuals with OSA with high accuracy. This is a useful approach that does not require any laborious analyses in a primary care setting or in remote areas where an initial specialized OSA diagnosis is not feasible. Support This study was supported in part by the Japan Society for the Promotion of Science (grant numbers 17K11793, 19K10236).

Download Full-text

Detection of obstructive sleep apnea using Belun Sleep Platform wearable with neural network-based algorithm and its combined use with STOP-Bang questionnaire

PLoS ONE ◽

10.1371/journal.pone.0258040 ◽

2021 ◽

Vol 16 (10) ◽

pp. e0258040

Author(s):

Eric Yeh ◽

Eileen Wong ◽

Chih-Wei Tsai ◽

Wenbo Gu ◽

Pai-Lien Chen ◽

...

Keyword(s):

Neural Network ◽

Obstructive Sleep Apnea ◽

Sleep Apnea ◽

Physiological Data ◽

Apnea Hypopnea Index ◽

Clinical Practices ◽

Sleep Study ◽

Link Type ◽

Obstructive Sleep ◽

Sensitivity Specificity

Many wearables allow physiological data acquisition in sleep and enable clinicians to assess sleep outside of sleep labs. Belun Sleep Platform (BSP) is a novel neural network-based home sleep apnea testing system utilizing a wearable ring device to detect obstructive sleep apnea (OSA). The objective of the study is to assess the performance of BSP for the evaluation of OSA. Subjects who take heart rate-affecting medications and those with non-arrhythmic comorbidities were included in this cohort. Polysomnography (PSG) studies were performed simultaneously with the Belun Ring in individuals who were referred to the sleep lab for an overnight sleep study. The sleep studies were manually scored using the American Academy of Sleep Medicine Scoring Manual (version 2.4) with 4% desaturation hypopnea criteria. A total of 78 subjects were recruited. Of these, 45% had AHI < 5; 18% had AHI 5–15; 19% had AHI 15–30; 18% had AHI ≥ 30. The Belun apnea-hypopnea index (bAHI) correlated well with the PSG-AHI (r = 0.888, P < 0.001). The Belun total sleep time (bTST) and PSG-TST had a high correlation coefficient (r = 0.967, P < 0.001). The accuracy, sensitivity, specificity in categorizing AHI ≥ 15 were 0.808 [95% CI, 0.703–0.888], 0.931 [95% CI, 0.772–0.992], and 0.735 [95% CI, 0.589–0.850], respectively. The use of beta-blocker/calcium-receptor antagonist and the presence of comorbidities did not negatively affect the sensitivity and specificity of BSP in predicting OSA. A diagnostic algorithm combining STOP-Bang cutoff of 5 and bAHI cutoff of 15 events/h demonstrated an accuracy, sensitivity, specificity of 0.938 [95% CI, 0.828–0.987], 0.944 [95% CI, 0.727–0.999], and 0.933 [95% CI, 0.779–0.992], respectively, for the diagnosis of moderate to severe OSA. BSP is a promising testing tool for OSA assessment and can potentially be incorporated into clinical practices for the identification of OSA. Trial registration: ClinicalTrial.org NCT03997916 https://clinicaltrials.gov/ct2/show/NCT03997916?term=belun+ring&draw=2&rank=1

Download Full-text

Application of machine learning to predict obstructive sleep apnea syndrome severity

Health Informatics Journal ◽

10.1177/1460458218824725 ◽

2019 ◽

Vol 26 (1) ◽

pp. 298-317 ◽

Cited By ~ 5

Author(s):

Corrado Mencar ◽

Crescenzio Gallo ◽

Marco Mantero ◽

Paolo Tarsia ◽

Giovanna E Carpagnano ◽

...

Keyword(s):

Machine Learning ◽

Obstructive Sleep Apnea ◽

Sleep Apnea ◽

Obstructive Sleep Apnea Syndrome ◽

Sleep Apnea Syndrome ◽

Support Vector ◽

Apnea Hypopnea Index ◽

Machine Learning Methods ◽

Obstructive Sleep ◽

Apnea Syndrome

Introduction: Obstructive sleep apnea syndrome has become an important public health concern. Polysomnography is traditionally considered an established and effective diagnostic tool providing information on the severity of obstructive sleep apnea syndrome and the degree of sleep fragmentation. However, the numerous steps in the polysomnography test to diagnose obstructive sleep apnea syndrome are costly and time consuming. This study aimed to test the efficacy and clinical applicability of different machine learning methods based on demographic information and questionnaire data to predict obstructive sleep apnea syndrome severity. Materials and methods: We collected data about demographic characteristics, spirometry values, gas exchange (PaO2, PaCO2) and symptoms (Epworth Sleepiness Scale, snoring, etc.) of 313 patients with previous diagnosis of obstructive sleep apnea syndrome. After principal component analysis, we selected 19 variables which were used for further preprocessing and to eventually train seven types of classification models and five types of regression models to evaluate the prediction ability of obstructive sleep apnea syndrome severity, represented either by class or by apnea–hypopnea index. All models are trained with an increasing number of features and the results are validated through stratified 10-fold cross validation. Results: Comparative results show the superiority of support vector machine and random forest models for classification, while support vector machine and linear regression are better suited to predict apnea–hypopnea index. Also, a limited number of features are enough to achieve the maximum predictive accuracy. The best average classification accuracy on test sets is 44.7 percent, with the same average sensitivity (recall). In only 5.7 percent of cases, a severe obstructive sleep apnea syndrome (class 4) is misclassified as mild (class 2). Regression results show a minimum achieved root mean squared error of 22.17. Conclusion: The problem of predicting apnea–hypopnea index or severity classes for obstructive sleep apnea syndrome is very difficult when using only data collected prior to polysomnography test. The results achieved with the available data suggest the use of machine learning methods as tools for providing patients with a priority level for polysomnography test, but they still cannot be used for automated diagnosis.

Download Full-text

Automated Detection of Obstructive Sleep Apnea Events from a Single-Lead Electrocardiogram Using a Convolutional Neural Network

Journal of Medical Systems ◽

10.1007/s10916-018-0963-0 ◽

2018 ◽

Vol 42 (6) ◽

Cited By ~ 23

Author(s):

Erdenebayar Urtnasan ◽

Jong-Uk Park ◽

Eun-Yeon Joo ◽

Kyoung-Joung Lee

Keyword(s):

Neural Network ◽

Obstructive Sleep Apnea ◽

Sleep Apnea ◽

Convolutional Neural Network ◽

Automated Detection ◽

Obstructive Sleep

Download Full-text

Convolutional Neural Network-Based Obstructive Sleep Apnea Identification

2021 IEEE 6th International Conference on Computer and Communication Systems (ICCCS) ◽

10.1109/icccs52626.2021.9449255 ◽

2021 ◽

Author(s):

Qimin Dong ◽

Yuttapong Jiraraksopakun ◽

Apichai Bhatranand

Keyword(s):

Neural Network ◽

Obstructive Sleep Apnea ◽

Sleep Apnea ◽

Convolutional Neural Network ◽

Obstructive Sleep

Download Full-text

Portable Sleep Apnea Syndrome Screening and Event Detection Using Long Short-Term Memory Recurrent Neural Network

Sensors ◽

10.3390/s20216067 ◽

2020 ◽

Vol 20 (21) ◽

pp. 6067

Author(s):

Hung-Chi Chang ◽

Hau-Tieng Wu ◽

Po-Chiun Huang ◽

Hsi-Pin Ma ◽

Yu-Lun Lo ◽

...

Keyword(s):

Neural Network ◽

Obstructive Sleep Apnea ◽

Sleep Apnea ◽

Recurrent Neural Network ◽

Short Term Memory ◽

Apnea Hypopnea Index ◽

Short Term ◽

Term Memory ◽

Obstructive Sleep ◽

Long Short Term Memory

Obstructive sleep apnea/hypopnea syndrome (OSAHS) is characterized by repeated airflow partial reduction or complete cessation due to upper airway collapse during sleep. OSAHS can induce frequent awake and intermittent hypoxia that is associated with hypertension and cardiovascular events. Full-channel Polysomnography (PSG) is the gold standard for diagnosing OSAHS; however, this PSG evaluation process is unsuitable for home screening. To solve this problem, a measuring module integrating abdominal and thoracic triaxial accelerometers, a pulsed oximeter (SpO2) and an electrocardiogram sensor was devised in this study. Moreover, a long short-term memory recurrent neural network model is proposed to classify four types of sleep breathing patterns, namely obstructive sleep apnea (OSA), central sleep apnea (CSA), hypopnea (HYP) events and normal breathing (NOR). The proposed algorithm not only reports the apnea-hypopnea index (AHI) through the acquired overnight signals but also identifies the occurrences of OSA, CSA, HYP and NOR, which assists in OSAHS diagnosis. In the clinical experiment with 115 participants, the performances of the proposed system and algorithm were compared with those of traditional expert interpretation based on PSG signals. The accuracy of AHI severity group classification was 89.3%, and the AHI difference for PSG expert interpretation was 5.0±4.5. The overall accuracy of detecting abnormal OSA, CSA and HYP events was 92.3%.

Download Full-text