Kinematic control differs in walking speed adjustment to different velocities

Abstract

[Purpose] To identify kinematic indices by performing acceleration/deceleration adjustment tasks and clarify the mechanism of walking speed adjustment. [Participants and Methods] Healthy adults with no history of orthopedic or central nervous system disease that could interfere with walking were included. Kinematic data with changes in walking speed were analyzed using a three-axis accelerometer, foot switches, and OptoJump Next. Two adjustment task experiments were conducted at different speeds: acceleration (Experiment 1) and deceleration (Experiment 2). The walking task constituted two conditions: walking at a comfortable speed and then shifting to the minimum speed as quickly as possible on a cue (minimum condition), and walking at a comfortable speed and then shifting to the intermediate speed as quickly as possible on a cue (intermediate condition). [Results] In Experiment 1, the step time and center-of-gravity acceleration in the front-back and left-right directions increased under the maximum condition for a longer period of time. In Experiment 2, the step length decreased earlier under the minimum condition; however, step time increased under the intermediate condition. [Conclusion] Kinematic control differs with adjustment to various target speeds. This study suggests that walking at a gait speed appropriate for a specific movement and environment can be improved through rehabilitation.

INTRODUCTION

Walking in daily life requires adjustments in walking speed according to a person’s purpose and environment^{1, 2)}. The ability to adjust walking speed decreases with aging and disease³⁾. Therapeutic strategies to improve the ability to adjust walking speed are crucial; however, effective methods have not yet been established. Conducting a detailed analysis of the underlying kinematic data variability during walking speed adjustment is crucial for establishing these strategies; however, limited studies have been conducted^{4, 5)}. Previous studies have only verified kinematic data under a certain limited condition owing to the use of the participant’s comfortable or maximum walking speed to unify the walking speed conditions^{4, 5)}. In contrast, walking speed adjustment in daily life shows a variety of patterns³⁾; nevertheless, the current methodology can only verify limited speed changes.

Rhythmic auditory cueing (RAC) is a walking practice method in which the participant walks in synchronization with sound, rendering it possible to set the participant’s walk at the desired speed by adjusting the tempo^6,7,8,9). Therefore, we deemed it possible to conduct walking practice with various speed changes by utilizing RAC and adjusting the tempo.

Based on the background, we aimed to clarify the characteristics of kinematic control during the adjustment of walking speed using RAC. Furthermore, we aimed to provide suggestions for rehabilitation strategies to improve the ability to adjust walking speed. The hypotheses of this study were that RAC can reproduce the adjustment task to the target walking speed and that kinematic control centered on step length is performed when the cadence is defined by the tempo.

PARTICIPANTS AND METHODS

Healthy adults who agreed to the purpose of the experiment and had no history of orthopedic or central nervous system diseases that could interfere with walking were included. Two experiments were conducted.

Experiment 1 included 17 participants (10 males and 7 females, mean age 24.1 ± 5.7 years), and Experiment 2 included 13 participants (5 males and 8 females, mean age 20.1 ± 0.3 years). All participants provided written informed consent. This study was approved by the Office of Research Ethics at Kanagawa University of Human Services (Approval No. 7-20-67) and conformed to the Declaration of Helsinki.

We used a 3-axis accelerometer (Noraxon, Scottsdale, AZ, USA), foot switches (Noraxon), and OptoJump Next (Microgate, Bolzano, Italy). The 3-axis accelerometer can record center-of-gravity acceleration during walking^{10, 11)}. It was affixed to the third lumbar spinous process, which is most reflective of center-of-gravity movement^{10, 11)}. The center-of-gravity accelerations were subjected to root mean square (RMS) processing, and the magnitudes of the center-of-gravity accelerations in the left-right (ax_RMS), front-back (ay_RMS), and vertical (az_RMS) directions were calculated^{12, 13)}. The sampling rate was 200 Hz. The foot switches were affixed to the center of both heels with tape and were used to identify the timing of heel contact during walking. The sampling rate was 2 kHz. OptoJump Next was installed on both sides of the walking path for an 8 m length, and the position of the grounded lower limb was recorded by infrared sensors. The sampling rate was 1 kHz^{14, 15)}. The kinematic data with walking speed changes were analyzed by synchronously recording these three types of devices.

A schematic of the acceleration-adjustment task is shown in Fig. 1. First, a 10 m walking test was performed three times each at the comfortable and maximum walking speeds to measure the walking speed used in the task. Participants were asked to “walk as fast as you feel comfortable walking” for the comfortable walking speed and “walk as fast as you can without running” for the maximum walking speed, and the average value for each speed was used as the representative value.

Fig. 1.

Experiment 1: acceleration task.

Next, the speed between the comfortable and maximum walking speed was calculated, and sounds were emitted from the speaker in rhythm with the cadence, allowing time to learn the intermediate walking speed. After the practice, the participants performed the 10 m walking test three times at the intermediate walking speed to confirm that they had learned the cadence. The walking task constituted two conditions: a task in which the participant started walking at a comfortable walking speed and then shifted to the maximum walking speed as quickly as possible on the cue (maximum condition), and a task in which the participant started walking at a comfortable walking speed and then shifted to the intermediate walking speed as quickly as possible on the cue (intermediate condition in Experiment 1). The order of trials for each condition was randomly determined, and a sufficient rest period was allocated between each condition. In both conditions, a computer program (LabVIEW 2015) was used to output a cue from the speaker at a random time. The analysis interval was one step before the cue and five steps after the cue. The RMS values of center-of-gravity acceleration (ax_RMS, ay_RMS, az_RMS), walking speed, step length, and step time were calculated for each step (from unilateral heel contact to contralateral heel contact) as the outcome measures.

A schematic of the deceleration-adjustment task is shown in Fig. 2. First, a 10 m walking test was performed three times each at the comfortable and the minimum walking speed to measure the walking speed used in the task. The minimum speed was set as “walking very slowly, as when looking at a painting in a museum”¹³⁾.

Fig. 2.

Experiment 2: deceleration task.

Next, the speed between the comfortable and maximum walking speeds was calculated for each participant. Sounds were emitted from the speaker with the cadence, allowing time to learn the tempo of the intermediate walking speed.

The walking task constituted two conditions: a task in which the participant started walking at a comfortable walking speed and then shifted to the minimum walking speed as quickly as possible on a cue (minimum condition), and a task in which the participant started walking at a comfortable walking speed and then shifted to the intermediate walking speed as quickly as possible on a cue (intermediate condition in Experiment 2). Five trials were conducted for each condition. The order of the trials for each condition was randomly determined, and a sufficient rest period was allocated between each condition. A computer program (LabVIEW 2015) was used to output a cue from the speaker at a random time. The analysis interval and the outcome measures were the same as those of Experiment 1.

Statistical analyses were performed using SPSS version 26.0 (IBM Corp., Armonk, NY, USA). Each measured item was analyzed using a linear mixed model with repeated measures. The fixed effects were the number of steps and the walking condition (Experiment 1: maximum and intermediate conditions; Experiment 2: minimum and intermediate conditions), whereas the random effects were the participants. The F- and p-values of the linear mixed model were calculated using the Type III test, and the Satterthwaite method was used to estimate the degrees of freedom. The restricted maximum likelihood method was used for parameter estimation. In addition, the number of steps was repeated measures, and the covariance structure of the repeated measures was first-order autoregressive. Multiple comparison tests (Holm’s method) were performed for items with significant differences. The same method was used for Experiments 1 and 2, and statistical significance for all the tests was set at 5%.

RESULTS

The results of Experiment 1 are shown in Figs. 3A and 4A.

Fig. 3.

Comparison between walking speed conditions based on kinematic parameters. A shows the results of the variation of walking speed, step time, and step length with walking conditions during the acceleration task (N=17, error bar: standard error). B shows the results of the variation of walking speed, step time, and step length with walking conditions during the deceleration task (N=13, error bar: standard error). The solid lines represent Comfortable to Max tasks. The dotted lines represent Comfortable to Intermediate tasks.

Fig. 4.

Comparison between walking speed conditions using root mean square (RMS). A shows the results of the variation of walking speed, step time, and step length with walking conditions during the acceleration task (N=17, error bar: standard error). B shows the results of the variation of walking speed, step time, and step length with walking conditions during the deceleration task (N=13, error bar: standard error). The solid lines represent Comfortable to Max tasks. The dotted lines represent Comfortable to Intermediate tasks.

Walking speed showed significant main effects and interactions for the number of steps and walking condition (number of steps: F_5,506=157.721, p<0.05; walking condition: F_1,102=148.184, p<0.05; interaction: F_5,436=17.589, p<0.05) (Fig. 3A). The walking speed significantly increased under the maximum condition compared to the intermediate condition from the first step, with a gradual increase until the third step. The walking speed gradually increased until the second step under the intermediate condition.

Step length showed significant main effects and interactions for the number of steps and walking condition (number of steps: F_5,411=38.725, p<0.05; walking condition: F_1,146=126.676, p<0.05; interaction: F_5,385=3.850, p<0.05). The step length significantly increased under the maximum condition compared to the intermediate condition for all steps, with a gradual increase until the second step. The step length significantly increased in the third and fourth steps compared to the first step under the intermediate condition.

Step time showed significant main effects and interactions for the number of steps and walking condition (number of steps: F_5,484=61.251, p<0.05; walking condition: F_1,110=78.430, p<0.05; interaction: F_5,430=4.430, p<0.05). The step time significantly decreased under the maximum condition compared to the intermediate condition from the first step, with a gradual decrease until the fourth step. The step time significantly decreased at the third and fourth steps compared to the second step under the intermediate condition.

ax_RMS showed significant main effects and interactions for the number of steps and walking condition (number of steps: F_5,401=56.740, p<0.05 walking condition: F_1,171=104.318, p<0.05; interaction: F_5,382=5.512, p<0.05) (Fig. 4A). ax_RMS significantly increased under the maximum condition compared to the intermediate condition from the first step, with a gradual increase until the fourth step. ax_RMS significantly increased from the first step under the intermediate condition compared to comfortable walking.

ay_RMS showed significant main effects and interactions for the number of steps and walking condition (number of steps: F_5,384=67.896, p<0.05; walking condition: F_1,189=103.365, p<0.05; interaction: F_5,370=7.915, p<0.05). ay_RMS significantly increased under the maximum condition compared to the intermediate condition from the second step, with a gradual increase until the third step. Furthermore, it significantly increased in the fifth step compared to the third step. ay_RMS significantly increased in all steps after the cue under the intermediate condition compared to comfortable walking.

az_RMS showed only significant main effects for the number of steps and walking condition (number of steps: F_5,370=3.607, p<0.05; walking condition: F_1,246=44.504, p<0.05). az_RMS significantly increased under the maximum condition compared to the intermediate condition for all steps, with a significant increase at the first and fifth steps compared to comfortable walking. az_RMS significantly increased in all steps after the cue under the intermediate condition compared to comfortable walking.

The results of Experiment 2 are shown in Figs. 3B and 4B.

Walking speed showed significant main effects and interactions for walking speed for the number of steps and walking condition (number of steps: F_5,509=323.366, p<0.05; walking condition: F_1,96=182.463, p<0.05; interaction: F_5,510=21.463, p<0.05) (Fig. 3B). walking speed significantly decreased under the minimum condition compared to the intermediate condition for all steps, with a gradual decrease until the third step. The walking speed decreased gradually under the intermediate condition until the second step, with a significant decrease in the fifth step compared to the second step.

Step length showed significant main effects and interactions for the number of steps and walking condition (number of steps: F_5,480=27.176, p<0.05; walking condition: F_1,179=199.942, p<0.05; interaction: F_5,484=6.338, p<0.05). Step length significantly decreased under the minimum condition compared to the intermediate condition for all steps, with a gradual decrease until the second step. The step length significantly decreased at the fourth and fifth steps under the intermediate condition compared to comfortable walking, with a significant decrease in the fourth and fifth steps compared to the first step.

Step time showed significant main effects and interactions for the number of steps and walking condition (number of steps: F_5,496=45.023, p<0.05; walking condition: F_1,117=143.283, p<0.05; interaction: F_5,497=10.858, p<0.05). Step time significantly increased under the minimum condition compared to the intermediate condition for all steps, with a significant increase from the third to the fifth steps compared to the first step. The step time significantly increased from the first step under the intermediate condition compared to comfortable walking.

ax_RMS showed significant main effects and interactions for the number of steps and walking condition (number of steps: F_5,580=11.535, p<0.05; walking condition: F_1,243=21.634, p<0.05; interaction: F_5,586=3.849, p<0.05) (Fig. 4B). ax_RMS significantly decreased under the minimum condition compared to the intermediate condition from the second to the fifth step, with a significant decrease in the fourth and fifth steps compared to comfortable walking. Furthermore, it significantly decreased in the second, fourth, and fifth steps compared to the first step. Moreover, it decreased significantly in the fourth step compared to the second and third steps. No significant changes were observed under the intermediate condition.

ay_RMS showed significant main effects and interactions for the number of steps and walking condition (number of steps: F_5,604=26.707, p<0.05; walking condition: F_1,128=4.711, p<0.05; interaction effect: F_5,605=2.685, p<0.05). ay_RMS significantly decreased under the minimum condition compared to the intermediate condition from the third to the fifth steps, with a significant decrease from the first steps compared to comfortable walking. Furthermore, it decreased significantly from the third to fifth steps compared to the first and second steps. ay_RMS significantly decreased from the second step after the cue under the intermediate condition compared to comfortable walking, with a significant decrease from the third to fifth step compared to the first step.

az_RMS showed a significant main effect only in the number of steps (F_5,602=5.105, p<0.05). az_RMS increased significantly under the minimum condition in the first, third, and fourth steps compared to comfortable walking. No significant changes were observed under the intermediate condition.

DISCUSSION

Step length increased until the second step under the maximum condition, whereas step time decreased until the fourth step. On the other hand, step length and step time increased and decreased, respectively, under the intermediate condition at the same time as the third and fourth steps. Previous studies have reported that step length increases with an increase in ankle plantar flexor muscle activity after the start of acceleration^{4, 16)}. These studies analyzed the kinematic data of a few steps of walking speed changes in a cross-sectional manner, whereas this study enabled a time-series analysis of the kinematic data using OptoJump Next. These findings suggest that control by step length may contribute earlier to a more substantial increase in walking speed for the same acceleration adjustment.

Regarding center-of-gravity acceleration, ax_RMS and ay_RMS increased stepwise under both conditions. Some studies have shown that the RMS of center-of-gravity acceleration depends on walking speed; therefore, the results of this study may likely reflect RMS changes depending on the walking speed^{17, 18)}. On the other hand, other studies have shown that changes in the joint angles of the hip and ankle contribute to acceleration by efficient absorption and release of energy^{5, 16)}. Therefore, verifying these indices in combination with kinematic data is necessary in the future.

In contrast, az_RMS slightly increased under the maximum condition, with no significant change observed under the intermediate condition. The vertical center-of-gravity acceleration is reported to increase with an increase in walking speed¹⁹⁾, and it is closely related to the energy efficiency of walking, with the amount of metabolizable energy increasing at both faster and slower speeds^{20, 21)}. In this study, a slight increase in az_RMS was observed, even under the maximum condition, which may be attributed to the fact that the vertical center-of-gravity acceleration is kept constant and does not fluctuate in healthy adults. These results suggest that the control is performed to suppress changes in az_RMS and to increase ax_RMS and ay_RMS in acceleration adjustment.

The step length decreased gradually until the second step under the maximum condition, and the step time increased until the third step. On the other hand, the step length began to decrease at the fourth step under the intermediate condition, and the step time increased from the first step. These results suggest that control by step length may be executed earlier with a greater decreasing speed and control by step time with a lower decreasing speed, even for the same walking speed deceleration adjustment. Few studies have investigated the adjustment of walking speed deceleration; nevertheless, some report that the adjustment of deceleration occurs from the walking stop task^{22,23,24,25,26)}. The muscle activity of the hip and knee extensors and ankle plantar flexors increases rapidly immediately in the early phase of the stance, decreasing the forward propulsive force²²⁾. In this study, the step length was significantly shortened from the first step under the minimum condition. This suggests that the walking speed may have been reduced by an early shortening of the step length in the case of a rapid decrease in walking speed, similar to the walking stop task. Regarding center-of-gravity acceleration, all axes showed a significant decrease compared to comfortable walking under the minimum condition; however, only ay_RMS showed a significant decrease under the intermediate condition. This suggests that ay_RMS may be an indicator for small changes in velocity and reflects the deceleration adjustments of increased lower limb muscle activity in the early phase. Deceleration adjustment from a comfortable walking speed was used in this study; therefore, the task difficulty was relatively low. Analyzing deceleration adjustments from various speeds and examining whether similar control occurs are necessary in the future.

This study had some limitations. Only two types of walking tasks were performed in this study; therefore, examining various patterns of speed changes and clarifying the changes in kinematic data remains necessary. In addition, this study included healthy participants, causing difficulty in generalizing the results to elderly individuals and patients. Nevertheless, the findings in this study serve as a basis for future analysis of kinematic control with reduced ability to adjust the walking speed. Furthermore, the length of the equipment used was limited, and we were only able to measure gait parameters over a limited distance. Given that different individuals, such as the elderly and patients, may require more distance to control their gait speed, we may need to conduct measurements over longer distances in the future.

The results of this study revealed that kinematic control during walking speed adjustment differs with different target speeds. Our findings suggested that it may be important to practice walking at a gait speed appropriate for the specific movement and environment targeted for improvement through rehabilitation.

Funding

This work was supported by the JSPS KAKENHI (grant number 23K10433).

Conflict of interest

The authors have no conflicts of interest, financial or otherwise, to declare.