The Influence of Recording Equipment on the Accuracy of Respiratory Rate Estimation from the Electrocardiogram and Photoplethysmogram

TY - CHAP

T1 - The Influence of Recording Equipment on the Accuracy of Respiratory Rate Estimation from the Electrocardiogram and Photoplethysmogram

AU - Charlton, Peter

AU - Bonnici, Tim

AU - Clifton, David

AU - Alastruey, Jordi

AU - Tarassenko, Lionel

AU - Beale, Richard

AU - Watkinson, Peter

PY - 2014

Y1 - 2014

N2 - line-height:13.7pt">INTRODUCTION: Respiratory rate (RR) is an important indicator of clinical deterioration, but is notoriously difficult to monitor in ambulatory patients. Algorithms have been proposed to derive RR from the electrocardiogram (ECG) and the photoplethysmogram (PPG), both of which can be obtained from ambulatory patients. Furthermore, these signals can be acquired wirelessly from commercial monitors for real-time RR estimation. However, commercial monitors apply filtering which may reduce the accuracy of RR estimation using these signals. Our aim was to determine whether routinely monitored ECG and PPG signals are suitable for accurate estimation of RR. line-height:13.7pt">METHODS. 58 healthy adults were recruited (42 under the age of 40 and 16 over the age of 70 years old). ECG and PPG were acquired simultaneously from high fidelity laboratory recording equipment configured to apply minimal filtering, a routine clinical monitor and a routine wireless pulse oximeter. RR estimates were obtained using time- and frequency-domain analyses of three respiratory modulations of the PPG and ECG. Their accuracy was assessed against a reference respiratory rate derived from electrical impedance plethysmography and oronasal airflow signals. line-height:13.7pt">RESULTS. The accuracy of RR estimates obtained from laboratory signals were significantly lower than those obtained from routine equipment for only one of the five estimation methods when applied to the PPG, and none when applied to the ECG. RR estimates obtained from elderly adults were less accurate than those obtained from young adults, raising concerns over the application of these algorithms in elderly patient populations. line-height:13.7pt">CONCLUSION. The filtering applied to ECG and PPG signals by commercial monitors does not affect the accuracy of RR estimation. However, further investigation is required to determine whether RR estimates can be obtained sufficiently accurately from elderly patients for routine use.

AB - line-height:13.7pt">INTRODUCTION: Respiratory rate (RR) is an important indicator of clinical deterioration, but is notoriously difficult to monitor in ambulatory patients. Algorithms have been proposed to derive RR from the electrocardiogram (ECG) and the photoplethysmogram (PPG), both of which can be obtained from ambulatory patients. Furthermore, these signals can be acquired wirelessly from commercial monitors for real-time RR estimation. However, commercial monitors apply filtering which may reduce the accuracy of RR estimation using these signals. Our aim was to determine whether routinely monitored ECG and PPG signals are suitable for accurate estimation of RR. line-height:13.7pt">METHODS. 58 healthy adults were recruited (42 under the age of 40 and 16 over the age of 70 years old). ECG and PPG were acquired simultaneously from high fidelity laboratory recording equipment configured to apply minimal filtering, a routine clinical monitor and a routine wireless pulse oximeter. RR estimates were obtained using time- and frequency-domain analyses of three respiratory modulations of the PPG and ECG. Their accuracy was assessed against a reference respiratory rate derived from electrical impedance plethysmography and oronasal airflow signals. line-height:13.7pt">RESULTS. The accuracy of RR estimates obtained from laboratory signals were significantly lower than those obtained from routine equipment for only one of the five estimation methods when applied to the PPG, and none when applied to the ECG. RR estimates obtained from elderly adults were less accurate than those obtained from young adults, raising concerns over the application of these algorithms in elderly patient populations. line-height:13.7pt">CONCLUSION. The filtering applied to ECG and PPG signals by commercial monitors does not affect the accuracy of RR estimation. However, further investigation is required to determine whether RR estimates can be obtained sufficiently accurately from elderly patients for routine use.

U2 - 10.5281/zenodo.439955

DO - 10.5281/zenodo.439955

M3 - Poster abstract

SN - 9780993039003

SP - 96

EP - 96

BT - MEC Annual Meeting and Bioengineering14 Programme and Abstracts

PB - MECbioeng14, Imperial College London

CY - London

ER -