Obstructive Sleep Apnea Detection from EEG Data: A Hybrid Approach of One-Dimensional Convolutional Neural Network and Enhanced Fuzzy C-Means Clustering Algorithm

Obstructive Sleep Apnea (OSA) is a serious sleep-breathing disorder often characterized by breathing interruptions causing life-threatening situation for the patient. The resource-intensive and expert-dependent nature of manual OSA detection techniques highlights the urgent demand for an automated OSA detection system. In this study, we propose an innovative approach that combines a One-Dimensional Convolutional Neural Network (1D-CNN) and Enhanced Fuzzy C-Means (E-FCM) Clustering Algorithm to automate OSA detection using EEG signal. By utilizing the 1D-CNN architecture, the data is subjected to training. Salient features are then taken from the pre-processed EEG data at 30-s epochs, and subsequently transformed into lower dimensional feature vectors. Following the feature extraction process, we seamlessly fed the resultant feature vectors into the E-FCM clustering algorithm for precise and efficient OSA detection. Our methodology achieved remarkable accuracy of 99.20 and 99.57% from two different EEG datasets. Our proposed methodology offers a promising solution for automated OSA detection, addressing a critical healthcare challenge.