Waveform-based interlimb ground-reaction-force similarity reveals direction-dependent locomotor adaptation during split-belt treadmill walking.

Split-belt walking has been widely used to identify the interlimb motor learning for decades. Although split-belt adaptation is commonly quantified using spatiotemporal variables or discrete peak kinetic measures, these observations cannot capture all aspects of similarity between right and left ground reaction force waveform patterns across the entire stance phase. This study examined (1) whether GRFSim, a measure of interlimb ground-reaction-force waveform similarity using normalized cross correlation (XCorr) could identify changes that occur during adaptation and de-adaptation while walking on split-belt treadmill, and (2) whether the detected changes were direction-dependent across anteroposterior (AP), mediolateral (ML), or vertical (V) GRF components. Twenty healthy young adults walked under four conditions: tied-belt fast walking (1.0 m/s), tied-belt slow walking (0.5 m/s), split-belt adaptation (left leg: 0.5 vs. right leg: 1.0 m/s), and tied-belt washout (0.5 m/s). Mean GRFSim and its stride-to-stride variability of GRFSim were calculated from the first or last 50 gait cycles of six periods: late fast baseline, late slow baseline, initial adaptation, late adaptation, initial de-adaptation, and late de-adaptation. Friedman tests with post-hoc comparisons using Wilcoxon signed-rank tests were used to assess GRFSim in aforementioned periods. GRFSim was lowest and its stride-to-stride variability was greatest during the initial adaptation (all p < 0.001). GRFSim remained lower than baseline during late adaptation and initial de-adaptation. The largest reduction in GRFSim occurred in the AP direction compared with the other directions, suggesting direction-dependent changes related to interlimb braking and propulsion. XCorr-derived GRFSim may complement spatiotemporal measures in locomotor-adaptation research.