Influence of low-pass filtering strategies on lower limb joint moments during walking

Joint moments derived from inverse dynamics are widely used to infer muscular demand during walking. While low-pass filtering of kinematic and kinetic data is essential, the choice of cutoff frequency combinations varies substantially and can influence joint moment estimates. Although these effects are well documented in dynamic sports movements, their impact during walking remains unclear. Therefore, the purpose of this study was to examine how different kinematic-kinetic low-pass filter cutoff combinations affect lower limb joint moments during level walking. Kinematic and kinetic data were filtered with a fourth-order Butterworth filter using four cutoff frequency pairings: 6-50, 6-6, 10-10 and 15-15 Hz. Peak hip extensor, hip abductor, knee extensor, and ankle extensor moments were compared, with waveform differences further assessed using one-dimensional statistical parametric mapping. Results showed that filter selection substantially affected peak hip extensor moments, which were reduced as the matched cutoff frequencies were lowered, differing up to two-fold between 15 and 15 and 6-6 Hz. Hip abductor and knee extensor moments showed smaller but significant differences, whereas the ankle extensor moment was unaffected. Filtering also altered hip extensor moment waveforms, particularly during early stance, with mismatched and higher cutoffs producing sharp, physiologically implausible moment spikes. Even during walking, joint moment estimates are highly sensitive to filtering choices, especially at the hip. Matched cutoff frequencies as low as 6-6 Hz may yield more physiologically reasonable hip extensor moment profiles. Transparent reporting of filtering practices is essential for interpretation and comparability of walking biomechanics studies.