Analysis of Breathing Rate Estimation Using Adaptive Techniques

Abstract

Breathing rate (BR) is an important physiological factor that is commonly measured in many treatment settings. Though it is still routinely measured by hand. In this study, a novel approach for determining the BR from an ECG, photoplethysmogram, or blood pressure signal is proposed. To extract respiratory signals from time and frequency domain data, the framework employs Discrete Wavelet Transform and Empirical Mode Decomposition techniques. Because we used a Robust Kalman Filter with a Signal Quality Index, our technique worked effectively even when the signals were severely damaged. The output signals have been integrated by state vector fusion, and the BR has been established. Two openly available clinical databases, the MIT-BIH Polysomnographic and the BIDMC datasets are used. The mean absolute percentage error was used to assess performance. The results were very accurate; PPG signals had MAPEs of 7% and BP signals of 5.4%, whereas ECG signals on the two databases had MAPEs of 4% and 4%, respectively.Additionally,the results revealed an astounding robustness to noise at 0 db. In light,this technique may be beneficial for BR monitoring in noisy areas.