A pulse transit time based fusion method for the noninvasive and continuous monitoring of respiratory rate

• Main Authors: Xiao-Rong Ding, Yuan-Ting Zhang, Hon Ki Tsang, Karlen, Walter
• Format: Conference Proceeding
• Language: English
• Subjects: Benchmark testing; Biomedical measurement; Electrocardiography; Estimation; Monitoring; Reliability; Temperature measurement
• ISSN: 1557-170X; 2694-0604
• DOI: 10.1109/EMBC.2016.7591663

The purpose of this study was to investigate whether pulse transit time (PTT) can be used for continuous monitoring of respiratory rate (RR). We derived PTT from the electrocardiogram and photoplethysmogram obtained from 42 recordings of CapnoBase, a publicly available benchmark data set for validating respiratory related measurements. The number of breaths in a minute (RR#) was estimated from the heart rate interval (HRI), pulse rate interval (PRI), and from PTT. In addition, to improve the estimation reliability, we investigated a fusion of the three HRI, PRI and PTT derived estimations. The root mean squared error (RSME) and a Bland-Altman plot were calculated using RR from capnography as reference. Finally, the proposed method was compared against the CapnoBase Smart Fusion RR benchmark estimation which estimates RR with three parameters extracted from the PPG signal alone. Thirty-seven recordings showed sufficient signal quality to estimate RR from PTT. The fused RR (RMSE 1.76 breaths/min) was more accurate than the estimations from PTT (RMSE 2.63 breaths/min), HRI (RMSE 1.96 breaths/min), and PRI (RMSE 2.73 breaths/min) alone. The proposed method also outperformed the CapnoBase benchmark (RMSE 3.08 breaths/min) algorithm. This study demonstrates that PTT is a valuable noninvasive parameter from which RR can be estimated.