Abstract—Sleep apnea, a serious sleep disorder characterized by interrupted breathing during sleep, poses significant health risks such as cardiovascular disease and diabetes. Traditional diagnostic methods, such as polysomnography, are cumbersome and expensive, creating a demand for automated solutions. Previous sleep apnea detection research relied on Multi-Layer Perceptrons (MLPs), which, while functional, can be limited in capturing the complex temporal dependencies within Electroencephalogram (EEG) signals. Our study introduces a novel approach by combining Convolutional Neural Network (CNN) with Fast Fourier Transform (FFT) to detect sleep apnea. The FFT was implemented as feature extraction to capture frequency-domain characteristics of the EEG signals. Two versions of the model were developed: one trained on raw EEG data and the other on FFT-processed data. The Physionet Sleep-EDF Database served as the source for EEG recordings, labeled as “normal” or indicative of sleep-disordered breathing. Performance metrics, including accuracy, precision, recall, and F1-Score, were used to evaluate the models. The CNN trained on raw EEG data achieved superior results, with 92% accuracy, a precision of 0.86, recall of 1.00, and an F1-Score of 0.92, outperforming previous studies utilizing Multi-Layer Perceptrons (MLP). However, the result shows that the approach using FFT produces worse results. This suggests that, in the context of sleep apnea detection using our specific dataset, the most discriminative features may not reside solely in the frequency domain as extracted by FFT. The results demonstrate the potential of CNNs in developing low-cost, accessible diagnostic tools. Future efforts should address dataset limitations and explore alternative feature extraction methods to improve generalizability.
 
Keywords—Convolutional Neural Network (CNN), deep learning, Electroencephalogram (EEG), sleep disorder, sleep apnea


Copyright © 2025 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).