Detection of Sleep Apnea From ECG Signal Using Various Machine Learning Algorithms

Abstract

The common fatal sleep disorder is sleep apnea (SA). This is demonstrated when a person's breathing is obstructed during sleep due to inadequate air deliver to the brain. Early recognition of sleep apnea will result in better outcomes from serious health issues. The precise procedure for detecting sleep apnea can improve the quality of life for individuals. Automatic Computer-Aided Detection (CAD) of sleep apnea was established, and it may effectively cut costs, and it is also used at home. The ECG signal is divided into Approximate and Detailed coefficients by utilizing DWT. The approximated and detailed coefficients are used to calculate the features such as Standard Deviation, Mean, Median, Energy, Maximum and Minimum Wavelet Coefficients. Machine learning algorithms like Naïve Bayes (NB), Support vector machine (SVM), and Decision Tree classifier (DT) are used to identify the sleep apnea. The accuracy (AC), recall, specificity (SP), precision, and F1-score of the developed technique employing the DT classifier were 98.07%, 97.85%, 98.4%, 98.89% and 98.32% respectively. This method outperformed prior methods using the same database and was shown to be more effective, powerful, and simple to use.