Recognizing the breathing resistances of wearing respirators from respiratory and sEMG signals with artificial neural networks

This study is devoted to recognizing the breathing resistances of wearing respirators from respiratory and surface electromyography (sEMG) signals. Ten subjects were required to sit for 5 min and walk for 5 min while wearing two different models of N95 filtering facepiece respirators (FFRs) and without a respirator. We recorded the sEMG signals from the respiratory muscles of the subjects, and the respiratory amplitude is also collected. Subsequently, fifteen features of the sEMG time domain and respiratory amplitude were extracted and used as input vectors to a recognition model based on artificial neural networks (ANNs). Finally, the experimental results show that these artificial neural networks are effective for recognizing different airway resistances of wearing respirators from sEMG and respiratory signals. The results also indicate that abdominal and scalene are the primary respiratory muscles affected by using N95 FFRs. Relevance to industry Respirator manufactures and administrations can readily employ this paper's findings for recognizing the breathing resistances of wearing respirators from respiratory and surface electromyography (sEMG) signals based on artificial neural networks automatically. Observations of the present study are in support of testing only the two primary muscles (abdominal and scalene) to simplify the evaluation of the effects of the breathing resistances of wearing respirators on respiratory muscles.