Development of a New Adaptive Method Based on Empirical Fourier Decomposition for the Diagnosis of Obstructive Sleep Apnea Using Electrocardiogram Signal Analysis

Obstructive sleep apnea (OSA) is a common sleep-related breathing disorder that can have significant effects on people's quality of life and daily functioning. polysomnography is the gold standard for diagnosing sleep apnea which cannot provide the expectations of a fast and economical diagnosis by analyzing several signals simultaneously. In this regard, the development of automatic, reliable and cost-effective diagnosis algorithms is important. Therefore, in this study, with the aim of diagnosing obstructive sleep apnea events, an automatic diagnostic algorithm based on single-lead Electrocardiogram (ECG) signal has been proposed. For this purpose, a new adaptive method based on Empirical Fourier Decomposition (EFD) and extraction of statistical and fractal dimension features from the Fourier Intrinsic Band Functions (FIBF) of the signal along with the ReliefF selection algorithm and Random Forest classification has been used. Empirical Fourier Decomposition can be a new tool for signal decomposition, which provides a suitable capability in extracting oscillations related to non-stationary components of the signal. In order to evaluate the proposed algorithm, the Apnea-ECG database, which contains 70 recordings of single-channel ECG signals, has been used. The results have shown that the proposed algorithm is able to detect obstructive sleep apnea events with %88.03 accuracy, %83.44 sensitivity, and %90.84 specificity. The high accuracy of the obtained results along with the number of suitable features indicates a compromise between the accuracy and the number of extracted features, which leads to a suitable computational load for the proposed algorithm, which makes it possible to use it in clinical applications.