Método de diagnóstico da síndrome da apneia obstrutiva do sono por aprendizado de máquina

Abstract

Obstructive sleep apnea syndrome (OSAS) is characterized by fragmentation and repetitive hypoxia during sleep, if this syndrome is not properly diagnosed and treated, it becomes the cause of serious complications such as cardiovascular problems. The diagnosis of this syndrome requires a detailed clinical examination called polysomnography, which consists of several tests that perform an analysis of brain (EEG), heart (ECG), muscle (EMG) and eye (EOG) activity. Due to the complexity of performing polysomnography, the present study aims to classify and diagnose two groups of subjects, healthy and with normal apnea, based on the use of ECG signals applied in a supervised machine learning algorithm along with Principal Component Analysis (PCA). Using the feature extraction methodology adapted for the diagnosis of obstructive sleep apnea, the results were sampled in two and three dimensions with 95% accuracy.