Effect of Sleep EEG, EOG, ECG and SpO2 Signals on Sleep Apnea Classification

Abstract

Sleep apnea is the cessation ofbreathing for at least 10 seconds during sleep. Sleep apnea is diagnosed by evaluating the signals recorded by the Polysomnography (PSG) Device by a specialist physician. However, this evaluation takes too much time for the physician. Detection of this evaluation by means of a software will reduce the workload of the relevant physicians and allow them to work more in different areas where needed. This study aims to reduce the workload of the physicians. For this purpose, features were extracted from EEG, ECG, EOG and SPO2 signals. Then, they were classified with Support Vector Machines (SVM) and K Nearest Neighbor Algorithm (KNN). The patient's signals were divided into epochs, and these epochs were labeled as apneic and healthy. Then, the classification was made and the correct results were compared with the predicted results. According to the results obtained, it was seen that the use of all EEG, ECG, EOG and SPO2 signals gave the highest accuracy rate in Support Vector Machines (SVM) Classification and the highest accuracy rate was determined as 92.5%. © 2024 IEEE.