FENet: A Frequency Extraction Network for Obstructive Sleep Apnea Detection

Abstract

Streszczenie-Obstructive Sleep Apnea (OSA) is a highly prevalent but inconspicuous disease that seriously jeopardizes the health of human beings. Polysomnography (PSG), the gold standard of detecting OSA, requires multiple specialized sensors for signal collection, hence patients have to physically visit hospitals and bear the costly treatment for a single detection. Recently, many single-sensor alternatives have been proposed to improve the cost efficiency and convenience. Among these methods, solutions based on RR-interval (i.e., the interval between two consecutive pulses) signals reach a satisfactory balance among comfort, portability and detection accuracy.In this paper, we advance RR-interval based OSA detection by considering its real-world practicality from energy perspectives. As photoplethysmogram (PPG) pulse sensors are commonly equipped on smart wrist-worn wearable devices (e.g., smart watches and wristbands), the energy efficiency of the detection model is crucial to fully support an overnight observation on patients. This creates challenges as the PPG sensors are unable to keep collecting continuous signals due to the limited battery capacity on smart wrist-worn devices. Therefore, we propose a novel Frequency Extraction Network (FENet), which can extract features from different frequency bands of the input RRinterval signals and generate continuous detection results with downsampled, discontinuous RR-interval signals. With the help of the one-to-multiple structure, FENet requires only one-third of the operation time of the PPG sensor, thus sharply cutting down the energy consumption and enabling overnight diagnosis. Experimental results on real OSA datasets reveal the state-ofthe-art performance of FENet.

Aquaculture systems in Hanoi, Vietnam.

“…Yang et al [26] then took this a step further by unveiling a one-dimensional squeeze-and-excitation (SE) residual group network that intricately captures the interplay between heart rate variability (HRV) and ECG-derived respiration (EDR). In another works, Chen et al [27] proposed the BAFNet. This innovative architecture employs a bottleneck attention-based fusion network, targeting key ECG parameters like R-R intervals and R-peak amplitudes, and synergizing convolutional networks with a global query method for a more precise identification of SA.…”

“…• Filtering During this stage, we utilize a Bandpass Finite Impulse Response (FIR) filter, following a similar approach as presented in previous works [12], [27]. This selection enables us to isolate significant signal features while eliminating frequencies beyond this range, thereby…”

Telemonitoring for the Follow-Up of Obstructive Sleep Apnea Patients Treated with CPAP: Accuracy and Impact on Therapy

“…The definitions and algorithms used for the automatic detection of residual AHI in CPAP are crucial to understanding the collected parameters. When measured by CPAP device, the definition of AHI is completely different from that used for assessment by polysomnography (PSG) [ 7 , 8 ].…”

Heterogeneous Collaborative Learning for Personalized Healthcare Analytics via Messenger Distillation

“…Specifically, the model will tell whether there is an apnea event based on a 60-dimensional RR-interval (time interval between two R-peak) vectors. Such classification pipelines greatly facilitate OSA patients' self-monitoring since they can get detection results via wireless wearable devices like smart-watches and smart-bands (Ye, Yin, Chen, Chen, Cui, and Zhang, 2021).…”