| Title |
A Study on Sleep Apnea Detection using Sleep Sound Data based on 1D-ResNet Deep Learning Model |
| Authors |
박태원(Park Tae-Won) ; 윤승원(Yoon Seung-Won) ; 윤혜원(Yoon Hye-Won) ; 이규철(Lee Kyu-Chul) |
| DOI |
https://doi.org/10.5573/ieie.2024.61.12.74 |
| Keywords |
Sleep apnea; Polysomnography; Sleep sound; Deep learning; MFCC |
| Abstract |
Sleep Apnea is a serious global health issue caused primarily by the obstruction of the upper airway during sleep. The current standard diagnostic method for sleep apnea, Polysomnography (PSG), has limitations such as high cost and complexity. In this study, we propose a non-invasive method to detect sleep apnea by utilizing sound data recorded during sleep. The sound data features were extracted using Mel-Frequency Cepstral Coefficients (MFCC), and classification was performed using a 1D CNN-based ResNet (Residual Network) model. The experimental results show that the proposed model achieved an average accuracy of 97.8%, a recall of 97.7%, a precision of 97.9%, and an AUC of 0.978 through 5-fold cross-validation. Future research will focus on expanding the dataset and experimenting with various deep learning models to further improve the model's performance. This study is expected to contribute to improving the accuracy of sleep apnea detection and the development of an efficient healthcare management system. |