Modulation of gait control under metronomic constraints assessed by long-range autocorrelations in healthy population and in Parkinson's disease

Background: Gait variability of healthy adults displays Long-Range Autocorrelations (LRA), meaning that the stride interval at any step statistically depends on remote previous gait cycles. Clinically, the presence of LRA in stride series is often considered as a marker of health, while its alteration has been associated with pathological conditions such as Parkinson’s disease. Previous works suggested that LRA may result from a selective regulation of stride length and duration while maintaining a constant target speed. In this framework, constraining the stride time with a metronome would impedes the flexible regulation of this gait parameter, leading to a decrease of LRA in series of stride durations. However, while the decrease in LRA when gait is guided with a metronome has been widely documented, transitions between walking with and without a metronome (and vice-versa) have not been measured. Methods: A total of 21 healthy volunteers were asked to walk overground during two conditions of 15 minutes in which a metronome was activated during either the first or second half of the session to test both transitions. Participants were explicitly asked to synchronize one foot with the metronome, set to their average spontaneous pace. The LRA was assessed with the Adaptive Fractal Analysis and this method was applied over a sliding window on the stride series to outline its evolution. Our analyses were reproduced with a computational model allowing to relate sudden changes in movement parameters to the LRA, which is typically measured over longer timescales. In the light of current theories associating the basal ganglia with a representation of motor costs, we are currently applying the same protocol involving patients with Parkinson’s disease and age-matched healthy adults in order to interpret their adaptation within the same framework. Results: In healthy participants, results showed a clear transition in both conditions, with LRA of the series of stride duration gradually reduced when the metronome was turned on and recovered when it was turned off. In contrast, the task instruction had little impact on the standard deviation of the same series. In both conditions, the change in LRA could be reproduced in the model by an instantaneous switching of the control policy, gait guided with a metronome being modelled by adding a term regulating asynchrony between the metronome beat and heel strike. The same analyses will be conducted on patients’ data and presented in detail. Preliminary results showed a significant decrease in LRA with the metronome, despite greater heterogeneity across patients compared to age-matched controls. Conclusions: Our results validate the hypothesis that LRA emerge from a flexible control strategy that rapidly regulates timing and amplitude parameters according to task requirements and provide a complementary methodology to investigate the effect of task instruction to assess gait in healthy and clinical population.