Abstract:A variety of feature extraction and classification approaches have been proposed using electrocardiogram (ECG) and ECG-derived signals for improving the performance of detecting apnea events and diagnosing patients with obstructive sleep apnea (OSA). The purpose of this study is to further evaluate whether the reduction of lower frequency P and T waves can increase the accuracy of the detection of apnea events. This study proposed filter bank decomposition to decompose the ECG signal into 15 subband signals, and a one-dimensional (1D) convolutional neural network (CNN) model independently cooperating with each subband to extract and classify the features of the given subband signal. One-minute ECG signals obtained from the MIT PhysioNet Apnea-ECG database were used to train the CNN models and test the accuracy of detecting apnea events for different subbands. The results show that the use of the newly selected subject-independent datasets can avoid the overestimation of the accuracy of the apnea event detection and can test the difference in the accuracy of different subbands. The frequency band of 31.25–37.5 Hz can achieve 100% per-recording accuracy with 85.8% per-minute accuracy using the newly selected subject-independent datasets and is recommended as a promising subband of ECG signals that can cooperate with the proposed 1D CNN model for the diagnosis of OSA.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: **Evaluating whether reducing low - frequency P - waves and T - waves can improve the accuracy of sleep apnea event detection**. Specifically, the researchers decomposed the electrocardiogram (ECG) signal into multiple sub - band signals through filter bank decomposition, and combined with a one - dimensional convolutional neural network (1D CNN) model to extract and classify the features of each sub - band signal, thereby evaluating the contribution of different frequency sub - bands to sleep apnea detection. ### Problem Background Obstructive sleep apnea (OSA) is a common sleep disorder, which is closely related to many health problems such as cardiovascular diseases. Traditional OSA diagnosis methods rely on polysomnography, but this method is expensive and time - consuming. In recent years, many studies have been devoted to developing OSA diagnosis systems based on single - lead electrocardiogram (ECG) signals to reduce costs and improve convenience. ### Research Objectives The main objective of this study is to evaluate **whether reducing low - frequency P - waves and T - waves can improve the accuracy of sleep apnea event detection**. Specifically, the researchers proposed the following hypotheses: - **Reducing low - frequency P - waves and T - waves** can enhance high - frequency R - waves, thereby highlighting the characteristics of RR intervals more clearly. - This improvement can enable the 1D CNN model to identify sleep apnea events more accurately. ### Methods To verify this hypothesis, the researchers adopted the following methods: 1. **Filter Bank Decomposition**: Use the Butterworth band - pass filter to decompose the original ECG signal into 15 sub - band signals. 2. **1D CNN Model**: Each sub - band signal is independently input into a 1D CNN model for feature extraction and classification. 3. **Dataset Selection**: Use 70 ECG records in the MIT PhysioNet Apnea - ECG database, which are divided into the original subject - dependent dataset and the newly selected subject - independent dataset. ### Key Findings The research results show that: - In the subject - independent test dataset, the **31.25 - 37.5 Hz sub - band signal** achieved an accuracy rate of 85.8% per minute and 100% per record. - Reducing low - frequency P - waves and T - waves does help to improve the accuracy of sleep apnea event detection, especially more obvious in the subject - independent dataset. ### Conclusions This study shows that the method of combining filter bank decomposition and 1D CNN model can improve the accuracy of sleep apnea event detection to a certain extent. In particular, the **31.25 - 37.5 Hz sub - band signal** is recommended as a potential ECG signal sub - band, which can be used in combination with the 1D CNN model for OSA diagnosis. ### Formula Summary The formulas involved in the article include: 1. **Z - score Normalization**: \[ z=\frac{x - \mu}{\sigma} \] where \(x\) is the filtered signal value, and \(\mu\) and \(\sigma\) are the mean and standard deviation of the filtered signal respectively. 2. **Performance Parameter Calculation**: - Accuracy (Accuracy, Acc): \[ \text{Acc}(\%)=\frac{\text{TP}+\text{TN}}{\text{TP}+\text{TN}+\text{FP}+\text{FN}}\times100\% \] - Sensitivity (Sensitivity, Sen): \[ \text{Sen}(\%)=\frac{\text{TP}}{\text{TP}+\text{FN}}\times100\% \] - Specificity (Specificity, Spec): \[ \text{Spec}(\%)=\frac{\text{TN}}{\text{TN}+\text{FP}}\times100\% \] These formulas are used for evaluation.

Contribution of Different Subbands of ECG in Sleep Apnea Detection Evaluated Using Filter Bank Decomposition and a Convolutional Neural Network

A Sleep Apnea Detection System Based on a One-Dimensional Deep Convolution Neural Network Model Using Single-Lead Electrocardiogram

A Multi-Scale Parallel Convolutional Neural Network For Automatic Sleep Apnea Detection Using Single-Channel Eeg Signals

Detection of obstructive sleep apnea from single-channel ECG signals using a CNN-transformer architecture

Obstructive Sleep Apnea Detection Using Ecg-Sensor with Convolutional Neural Networks.

Detection of Obstructive Sleep Apnoea Using Features Extracted from Segmented Time-Series ECG Signals Using a One Dimensional Convolutional Neural Network

Wavelet transform and deep learning-based obstructive sleep apnea detection from single-lead ECG signals

Obstructive sleep apnea prediction from electrocardiogram scalograms and spectrograms using convolutional neural networks

[Sleep Apnea Automatic Detection Method Based on Convolutional Neural Network].

A robust deep learning system for screening of obstructive sleep apnea using T-F spectrum of ECG signals

Single Channel ECG for Obstructive Sleep Apnea Severity Detection using a Deep Learning Approach

Multiclass classification of obstructive sleep apnea/hypopnea based on a convolutional neural network from a single-lead electrocardiogram

A Method to Detect Sleep Apnea Based on Deep Neural Network and Hidden Markov Model Using Single-Lead ECG Signal

Automatic detection of obstructive sleep apnea through nonlinear dynamics of single-lead ECG signals

Classification of sleep apnea based on EEG sub-band signal characteristics

Multi-Feature Automatic Extraction for Detecting Obstructive Sleep Apnea Based on Single-Lead Electrocardiography Signals

Recurrent Neural Network Based Classification of ECG Signal Features for Obstruction of Sleep Apnea Detection

Sleep Apnea Detection From Single-Lead ECG: A Comprehensive Analysis of Machine Learning and Deep Learning Algorithms

Multiscale Deep Neural Network for Obstructive Sleep Apnea Detection Using RR Interval From Single-Lead ECG Signal

Sleep Apnea Classification Algorithm Development Using a Machine-Learning Framework and Bag-of-Features Derived from Electrocardiogram Spectrograms

Single-lead ECG Based Multiscale Neural Network for Obstructive Sleep Apnea Detection.