Extraction of respiratory signals and respiratory rates from the photoplethysmogram

Abstract

Respiration rate (RR) is an important indicator of human health assessment which can be estimated by extracting respiratory signals from the photoplethysmogram (PPG). The goal of this study is to propose an alternative method, for obtaining accurate estimation of respiratory rate (RR) from the PPG signal. The proposed algorithm is based on the multiple autoregressive models and autocorrelation analysis (AC-AR). In AC-AR, the autoregressive model (AR) is applied to determining the dominant respiratory rate from the PPG, and autocorrelation is applied to reduce the effect of clutter in the three respiratory-induced variations. Meanwhile, this paper introduced signal quality indices (SQI) to improve reliability of results. This algorithm is tested using an open source database: The CapnoBase benchmark dataset, which comprising 42 eight-minute PPG recording and respiratory signal acquired form both children and adults in different clinical setting. Compared with that of existing method in the literature, the average absolute error percentage (AAEP) of the proposed algorithm is less than 3.72%, which demonstrated that our presented AC-AR bring a significant improvement in accuracy.