The Effect of Wavelet-based Filtering and Data Set Length on the Fractal Scaling of Cardiorespiratory Variability

The effect of filtering and data set length on the accuracy of the quantification of fractal characteristics of cardiorespiratory activity remains unclear. Breath-to-breath interval (BBI) and heartbeat-to-heartbeat interval (RRI) were recorded from 8 healthy human subjects during a quiet seated posture. Movement artifact was filtered from the raw respiratory data using a simple low-pass (LP) or a wavelet-based (WB) filter. The RRI data was segmented into three sets of 256, 512, and 1024 sequential data points. BBI and RRI fractal scaling was quantified using detrended fluctuation analysis and a wavelet-based estimation of fractal dimension. No significant difference in the calculation of fractal behavior of BBI was identified after using a LP or a WB filter. Furthermore, there was no significant difference in fractal measurements among the different RRI data set lengths. In conclusion, filtering of physiologic data with standard LP or WB techniques or data set length, between 256 and 1024 sequential points, does not significantly affect the calculation of fractal behavior.