Respiratory Rate Estimations using Three Respiratory-Induced Variations on Photoplethysmogram

Other than SpO2 and heart rate (HR), respiratory rate (RR) has become one of the baseline clinical characteristics as risk factors for Covid-19 mortality. Commonly measured using capnography and impedance pneumography (IP) in hospitals, respiratory rate is possibly incorporated in one portable device of pulse oximeter for home monitoring together with SpO2 and HR. Photoplethysmogram (PPG) signals, recorded by pulse oximeters, are modulated by respiration due to physiological mechanisms. There are three respiratory-induced variations on PPG signals: baseline wander (BW), amplitude modulation (AM), and frequency modulation (FM). From these three modulating signals, we estimated an RR value by various data fusions. In addition to the artifact handling mechanism, we also presented the effects of signal lengths on the performance of RR estimation methods in terms of error metrics and their ability to detect sudden changes in RR values.