A 1D-CNN Based Deep Learning Technique for Sleep Apnea Detection in IoT Sensors

Abstract

Internet of Things (IoT) enabled wearable sensors for health monitoring are widely used to reduce the cost of personal healthcare and improve quality of life. The sleep apneahypopnea syndrome, characterized by the abnormal reduction or pause in breathing, greatly affects the quality of sleep of an individual. This paper introduces a novel method for apnea detection (pause in breathing) from electrocardiogram (ECG) signals obtained from wearable devices. The novelty stems from the high resolution of apnea detection on a second-by-second basis, and this is achieved using a 1-dimensional convolutional neural network for feature extraction and detection of sleep apnea events. The proposed method exhibits an accuracy of 99.56% and a sensitivity of 96.05%. This model outperforms several lower resolution state-of-the-art apnea detection methods. The complexity of the proposed model is analyzed. We also analyze the feasibility of model pruning and binarization to reduce the resource requirements on a wearable IoT device. The pruned model with 80% sparsity exhibited an accuracy of 97.34% and a sensitivity of 86.48%. The binarized model exhibited an accuracy of 75.59% and sensitivity of 63.23%. The performance of low complexity patient-specific models derived from the generic model is also studied to analyze the feasibility of retraining existing models to fit patient-specific requirements. The patient-specific models on average exhibited an accuracy of 97.79% and sensitivity of 92.23%. The source code for this work is made publicly available.

A Deep Neural Network Based Wake-After-Sleep-Onset Time Aware Sleep Apnea Severity Estimation Scheme Using Single-Lead ECG Data

“…Kim et al [35] explored various machine learning models such as Logistic Regression, Random Forest, XGBoost as well as SVM and concluded that SVM showed the best result in their study. Many others imported assorted deep learning techniques such as various adaptations of CNN [7], [15], [16], [19], [38] or other deep learning models [11], [37] in an attempt to classify patients with OSA. Others used different biosignals such as SpO2 [39] or respiration signals such as Oronasal thermal airflow (FlowTh), Nasal pressure (NPRE), and abdominal respiratory inductance plethysmography (ABD) [40].…”

Machine Learning for Microcontroller-Class Hardware: A Review

“…TinyML opens up a broad spectrum of real-time and lowfootprint eHealth applications, some of which are summarized in Table XVI. These include monitoring eating episodes and coughs using microphones [221], [223], sleep monitoring and arrhythmia detection through ECG measurements [171], [222], epileptic seizure recognition from EEG sensors [171], and fall detection using earable inertial sensors [2]. Most TinyML mHealth applications are variants of anomaly detection, indicating the presence or the absence of a health condition, thereby allowing the use of ultralightweight models in the order of 10 0 -10 1 kB.…”

“…Most TinyML mHealth applications are variants of anomaly detection, indicating the presence or the absence of a health condition, thereby allowing the use of ultralightweight models in the order of 10 0 -10 1 kB. Example models for mHealth include Bonsai [2], embedded GRU [221], 1-D CNN [222], FC-AE [171], and two-layer CNN/LSTM [223].…”

A spatio-temporal learning-based model for sleep apnea detection using single-lead ECG signals

“…Following previous studies (Mostafa et al, 2017(Mostafa et al, , 2020, we labeled a 1 min segment as SA if the segment contains more than 5 s of SA events. Considering the class imbalance problem of UCDDB, the data of patients without SA events (ucddb008, ucddb011, ucddb013, and ucddb018) are not used (John et al, 2021).…”