An Artificial Intelligence-Based Approach for Automated Classification of Obstructive Sleep Apnea by Considering Multi-modal Feature Fusion Technique

Obstructive Sleep Apnea (OSA) is a crucial sleep-breath disorder often characterized by partial or complete cessation of airflow during sleep. The syndrome has a hazard effect on other physiological functions causing primary risk of fragmented sleep which is a part of Sleep Apnea Syndrome (SAS), headache and morning sickness. As secondary risks, there is a high chance of vehicular accidents, cardiac failure and stroke. For the diagnosis of OSA, polysomnography (PSG) is considered as the gold standard strategy that records and analyzes brain waves, heart rate, breathing pattern, oxygen level and artifacts of the survival. This paper uses a multi-modal approach by considering both ECG and SpO2 signals. Feature extractions of both the signals are carried out to extract time–frequency domain features and spatial features. Then, the extracted features are combined at a feature-level fusion technique. Finally, the fused features are fed to a series of machine learning classifiers such as Decision Tree (DT), Random Forest (RF) and Radial Basis Function-based Support Vector Machine (RBF-SVM). Comparing the classification performances, accuracy of RBF-based SVM outperformed other two classifiers with a score of 98.60%. Therefore, our proposed methodology can be considered for automated classification of OSA and non-OSA subjects. © 2023, The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd.