Last Updated on February 8, 2023 by pg@petergamma.org
If we choose a device which detects every beak of a respiration signal, then we should have a very accurate respiration rate. If we detect every ECG peak of the heart rate, and use it for a calibrated respiratory rate, should the accuracy then not also be very high? We do not know of a gold standard for respiratory rate measurement. If it is about life or death in the hospital, does some do a respiratory rate estimation to find out whether the patient is still alive or already dead? But to study the effect of practicing mediation on breath rate, for instance with breath meditation, it is not about life of dead. There are expensive devices on the market with a face mask on the market to measure respiration rate very accurately. But is this worth it for our own little private project?
The accuracy of the respiratory rate estimation of the Sensors paper with a Polar H10 chest strap was in the range of the accuracy of sports sensors. If we choose a 3 lead ECG with 3 measurements which are independent for each other, the accuracy increases, and we suppose therefore also the accuracy of respiratory rate estimaton. Garmin Fenix 6 or a higher model can measure respiration rate. It works with Garmin chest straps, but also Polar chest straps, and the Polar H10. If we use a 3 lead ECG device, the accuracy should increase. According to Peter Charlton, his respiratory rate estimation works for clinical settings. And there are dataset to calibrate the respiration rate. We suppose that the respiration rate on Garmin watches is based on respiratory rate, but the data are not calibrated. The YouTube video of Talha Iqbal with a Ph.D. from School of Medicine of the National University of Ireland show respiratory rate from PPG signal, which is less accurate, and uses Python. Respiratory rate from PPG is not very accurtate. We saw papers from fingerpletismographs which had an accuracy for heart rate which was about 80 %.
We are physiologists and not Biomedical Engineers specialising in signal processing for wearables, as Peter Charlton from the University or Cambridge is, and our knowledge about respiration rate accuracy is limited.