Investigating Genioglossus Electromyograms of Obstructive Sleep Apnea Patients and Healthy Subjects During...

Obstructive sleep apnea syndrome (OSAS) is a prevalent source of sleep-disordered breathing that is generally caused by common factors in modern life. OSA is most commonly associated with dysfunctions in the genioglossus. Thus, we aim to develop a genioglossus electromyogram (GG-EMG) mouthpiece connected to a device for diagnosis. This work presents the first part of the process: reading the GG-EMG. We use 10-second readings from 10 healthy subjects and 9 OSA patients aged 24-83 years during rest and activity. We then extract 37 features from each EMG, perform a Wilcoxon rank sum test, and extract Pearson's correlation coefficients (PCC) between the features to compare them at rest versus activity and control versus OSA. Afterward, we use continuous wavelet transform to get scalograms of the EMGs and compare extracted image features. We found that EMG features generally show a statistically significant difference (p < 0.05) and no correlation (|PCC| < 0.3) between OSA and Control subjects during both rest and active states, more with the latter than the former. We also found more statistically significant differences (p < 0.05) between subject image features in the rest state than the active state, meaning that it may be more convenient to skip the 10-second active recording during wakefulness.Significance- This work establishes initial testing of the mouthpiece with wakeful subjects, with the differences obtained showing that developing the mouthpiece is worthwhile.