ECG-Based Sleep Apnea Detection Using Convolutional and Long Short-Term Memory Networks

Sleep apnea, which is one of the most common sleep disorders that can lead to undesirable health outcomes if not treated and requires PSG as a standard diagnostic test for the diagnosis. Among the physiological parameters monitored by PSG during sleep are brain activity eye movement muscle activities heart rate and airflow. All of this makes it possible to identify apnea periods in sleep, which is a key symptom for sleep apnea diagnosis. Although PSG is acknowledged as a gold standard, there are other research interests in non-invasive diagnostic tools. One of such methods focuses on the possible role a single-signal ECG may play in sleep apnea diagnosis. One of the most important tools for measuring heart activity is ECG, giving useful information on what happens in a human being's heart. The suggested method is the segmentation of ECG signals and then feature extraction from them. This means that we might even be able to find sleep apnea markers without the intense long-term PSG monitoring. Lastly, the accuracy of our model at testing phase was determined to be 86.18% that suggests it an easy-to-use and effective diagnostic tool for sleep apnea cases.