Photoplethysmographic determination of the respiratory rate in acutely ill patients: validation of a new algorithm and implementation into a biomedical device

Abstract

BackgroundRespiratory rate is among the first vital signs to change in deteriorating patients. The aims of this study were to evaluate the accuracy of respiratory rate measurements using a specifically dedicated reflection-mode photoplethysmographic signal analysis in a pathological condition (PPG-RR) and to validate its implementation within medical devices.MethodsThis study is derived from a data mining project, including all consecutive patients admitted to our ICU (ReaSTOC study, ClinicalTrials.gov identifier: NCT02893462). During the evaluation phase of the algorithm, PPG-RR calculations were retrospectively performed on PPG waveforms extracted from the data warehouse and compared with RR reference values. During the prospective phase, PPG-RR calculations were automatically and continuously performed using a dedicated device (FreeO2, Oxynov, Québec, QC, Canada). In all phases, reference RR was measured continuously using electrical thoracic impedance and chronometric evaluation (Manual-RR) over a 30-s period.ResultsIn total, 201 ICU patients’ recordings (SAPS II 51.7 ± 34.6) were analysed during the retrospective evaluation phase, most of them being admitted for a respiratory failure and requiring invasive mechanical ventilation. PPG-RR determination was available in 95.5% cases, similar to reference (22 ± 4 vs. 22 ± 5 c/min, respectively; p = 1), and well correlated with reference values (R = 0.952; p < 0.0001), with a low bias (0.1 b/min) and deviation (± 3.5 b/min). Prospective estimation of the PPG-RR on 30 ICU patients’ recordings was well correlated with the reference method (Manual-RR; r = 0.78; p < 0.001). Comparison of the methods depicted a low bias (0.5 b/min) and acceptable deviation (< ± 5.5 b/min).ConclusionAccording to our results, PPG-RR is an interesting approach for ventilation monitoring, as this technique would make simultaneous monitoring of respiratory rate and arterial oxygen saturation possible, thus minimizing the number of sensors attached to the patient.Trial registry number ClinicalTrials.gov identifier NCT02893462

Usual and Advanced Monitoring in Patients Receiving Oxygen Therapy

“…In daily practice, this parameter may not be adequately recorded. 18 Recently, L'Her et al 19 demonstrated that breathing frequency could be obtained from photoplethysmography tracings. The breathing frequency derived from pulse oximetry signals correlated very well with breathing frequency measured manually, even in subjects with atrial fibrillation.…”

“…The breathing frequency derived from pulse oximetry signals correlated very well with breathing frequency measured manually, even in subjects with atrial fibrillation. 19 This function has been incorporated into some automated oxygen titration and weaning systems, enabling the continuous measurement of breathing frequency in patients wearing a pulse oximetry device. 19…”

“…19 This function has been incorporated into some automated oxygen titration and weaning systems, enabling the continuous measurement of breathing frequency in patients wearing a pulse oximetry device. 19…”

“…Breathing frequency can be continuously measured with a specific algorithm based on photoplethysmographic signal analysis. 19 Heart rate variability can also be measured with the photoplethysmographic signal analysis. 19 With automated oxygen titration, it is possible to combine these measured parameters in a new EWS that is focused on patients on oxygen therapy (Early Warning ScoreO 2 ) ( F3, F4 Fig.…”

A Contactless Respiratory Rate Estimation Method Using a Hermite Magnification Technique and Convolutional Neural Networks

“…Thermal images [9][10][11] measure the temperature variations between inhalation and exhalation phases with it not working if the nasal area is not visible. In recent years, several works have used the photoplethysmography technique, employed initially to measure cardiac frequency [12], to measures skin blood changes to track RR [13][14][15][16][17][18]. Some works such as [18] train a CNN using respiratory raw signal as reference and a skin reflection model to represent the color variations of the image sequence as input.…”