Analysis of Heart Rate Variation Filtering Using LMS Based Adaptive Systems

Heart Rate Variability (HRV) is widely used as an index of human autonomic nervous activity. HRV is composed of two major components: high frequency respiratory sinus arrhythmia (RSA) and low frequency sympathetic components. The ratio of LF/HF is viewed as an index of human autonomic balance, so the low frequency sympathetic and the high frequency parasympathetic components of an ECG R-R interval must be adequately separated. Adaptive filters can isolate the low frequency, enabling the attainment of more accurate heart rate variability measures. For the raised case, this paper suggests an efficient (short size) case based model and illustrates its performance in adaptive filtering of heart rate signal. This method renders analogous results to what a higher order conventional FIR model adaptive filter may yield. The strength of the proposed model comes out of its ability in tracking the phase difference variation between the reference and the main signal of an adaptive filtering system. This capability, then is shown, that leads to the increase in the convergence rate of the LMS algorithm in HRV adaptive filtering. Simulation results supporting the proposed concept are presented. Key-Words: Adaptive filter, All pass filter, FIR model, First order equalizer, HRV filtering, Rate of convergence, Least Mean Squares.