Using a Breakpoint to Determine the Optimal Cut-off Frequency

The aims of this study were to compare methods of determining the optimal cut-off frequency (CFopt) for a Butterworth filter. CFopt were determined for leg displacement data for treadmill running through residual analysis using regression (RA0reg), integral of the power spectral density (PSD), and both these methods analysed through a new ‘breakpoint’ method. RA0reg did not correlate with other methods suggesting poor concurrent validity. The ‘breakpoint’ method correlated significantly between several methods. CFopt was least for anterio-posterior and highest for vertical directions for all methods (p<0.05). Settings for RA0reg and PSD can have substantial effects on CFopt, but the ‘breakpoint’ is not affected as much by the settings. Future research should attempt to standardise settings and explore the criterion validity of the methods to determine CFopt.