Correlation Analysis Between the Estimated Respiratory Rate and Oxygen Saturation in Abnormal Respiratory Patterns

Recently, monitoring oxygen saturation (SpO2) has become necessary for the diagnosis of obstructive sleep apnea. Prior studies have proposed algorithms using radar sensors to estimate SpO2 by extracting vital signals such as chest displacement, respiratory rate (RR), and cardiac rate. However, the estimates obtained using these methods differ from the ground truth (GT) of SpO2 in abnormal respiratory patterns. To address these challenges, this study analyzed the relationship between the estimated RR and the GT. Several experimental results for three adults demonstrated that a time delay existed, with an average (avg) of 21.20 s and a standard deviation (SD) of ±6.7 s, to maximize the correlation coefficient between the RR and the GT. When stopping breathing, the GTs were decreased along the gradient with an avg of −0.036 s and SD of ±0.199 s. When restarting breathing, the GTs were increased along the gradient with an avg of 0.355 s and SD of ±0.399 s. By exploiting the analyzed results, we developed an algorithm to estimate the SpO2 in abnormal respiratory patterns.