The coordination of upper and lower limbs in curve-turning walking of healthy preschoolers: Viewed in continuous relative phase

Gait & Posture 75 (2020) 1–7

Contents lists available at ScienceDirect

Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost

The coordination of upper and lower limbs in curve-turning walking of healthy preschoolers: Viewed in continuous relative phase Quting Huanga, Mingyu Hua, Bo Xua, Jin Zhoua,b, a b

T

⁎

National Engineering Laboratory for Clean Technology of Leather Manufacture, Sichuan University, Chengdu 610065, PR China Science Lab, Zhejiang Red Dragonfly Footwear Co., LTD., Zhejiang Province, Wenzhou 325100, PR China

A R T I C LE I N FO

A B S T R A C T

Keywords: Kinematics Limb coordination Curve turning Preschooler Continuous relative phase

Background: Coordination is the ability to assemble and maintain appropriate relations between joints. Investigating limb coordination in curve-turning (CT) walking could provide insightful information about the coordinating strategies and neuro-musculoskeletal system (NMSS) control in human motion. Research question: Although preschoolers have already established an adult-like gait, how preschoolers perform their specific gait pattern when walking in CT and what coordination strategies they would choose during the turning process have not yet been systematically considered. Therefore, this study was aimed to investigate preschoolers' coordination mechanism during asymmetric motion, in order to understand the development of their NMSS control in locomotion. Methods: Kinematics data in the lower and upper limbs of 45 healthy preschoolers walking with the curveturning task was measured by the Coda Motion System. The Continuous Relative Phase (CRP) angle and the variability between the knee and ankle, hip and knee, as well as the thorax-humerus joint (THJ) and elbow were calculated. Results: The outcome demonstrates that as the curve angles increased, the stride length and Froude number of preschoolers significantly decreased (p < 0.05 for all); meanwhile, a more out of phase coordination pattern in CRP and an increase in VCRP values were found. Group analysis showed that the significant differences in CRP and VCRP between preschoolers and adults increased with curve angles in all coupled joints - the highest in that of the Knee-Ankle coupling, followed by those of the Hip-Knee and THJ-Elbow. Significance: Our results suggest that to achieve curve-turning, preschoolers first chose to modify their STP, then to adjust coordination for coupling-joints in the Knee-Ankle, Hip-Knee, and THJ-Elbow systems. Additionally, preschoolers are still in a gait fine-tuning period and their NMSS control of motion is not as precise as that of adults.

1. Introduction Human motion works via the Neuro-Musculo-Skeletal system (NMSS), and this macro-process can be observed and recorded; however, the inner relations between parts of this system, as well as its coordination mechanism are difficult to study directly. Since it has been suggested that coordination is an ability of the spinal neural circuitry to assemble and maintain appropriate relations between joints or segments while posture changes, studies of coordination would give insight into the role of NMSS control in human motion [1,2]. Meanwhile, the curve-turning (CT) task requires an advanced degree of development in motion, as it involves asymmetrical coordination in both the upper and lower limbs [3,4]. Preschoolers usually walk with an adult-like gait [5]; however, it is unknown whether their ability to synergize movement in ⁎

their upper and lower limbs is mature or still being fine-tuned. Hence, by performing CT tasks, we would be able to quantify differences between preschoolers and adults in motion strategies; and such findings would be beneficial for understanding preschoolers' development and NMSS control in locomotion. Kinematics and Coordination features of CT have been extensively reported. As for turning curves, the most obvious features for adult participants were a reduction in stride length, and their walking velocity [6,7]. Courtine et al. [3,4] systematically elucidated the CT mechanism in adults: participants first turned their head in the direction of curved path in advance once they perceived the task; then, they rotated their upper limb, trunk, and hips in the target direction; meanwhile, the trailing leg leaned outward to tilt the body inward; finally, they used the forces generated in the push-off of the support foot to accelerate the

Corresponding author at: National Engineering Laboratory for Clean Technology of Leather Manufacture, Sichuan University, Chengdu 610065, PR China. E-mail address: [email protected] (J. Zhou).

https://doi.org/10.1016/j.gaitpost.2019.09.013 Received 19 March 2019; Received in revised form 10 September 2019; Accepted 13 September 2019 0966-6362/ © 2019 Elsevier B.V. All rights reserved.

Gait & Posture 75 (2020) 1–7

Q. Huang, et al.

were marked on the floor (0°- 30°-60°) (Fig. 2). A three to five minutes' warm-up was first provided to participants; then they were asked to perform randomly-assigned curve-turning tasks. All participants were instructed to perform a step-by-step turn. One complete stride was defined as the period from initial right foot contact before a direction change to the subsequent right foot contact after a direction change. At least five successful strides were required in each measure. Eventually, after filtering out incomplete measurements, data of 33 preschoolers and 9 adults were selected in this study.

turning speed. In terms of coordination, Forsell et al. [8] reported that reduction of walking velocity would affect the axial coordination of turning. Patla et al. [9] concluded that control and shifting of the center of mass in the correct direction played an important role in successful turning, as the foot pushes off and the trunk rotates. Conversely, Courtine et al. [10] found no obvious association between the change of coordination patterns and curved walking, although kinematic adaptations occurred in lower limbs. Similarly, Grasso et al. [11] suggested that visual conditions had little effect on eye-head coordination during CT. Unfortunately, questions such as how preschoolers performed their specific gait pattern when turning curves and what coordination strategies they would choose during the turning process have not yet been systematically considered. According to current knowledge, the continuous relative phase (CRP) is a simple and practical protocol to investigate coordination in various motion conditions [1,2,12–14]. The CRP is calculated based on the spatial and temporal parameters of the involved joints and contains both spatial and temporal information of these two joints. These variables provide both relative motion between two joints and a higher resolution form of the discrete relative phase for assessing coordination. Furthermore, Chiu et al. [2] indicated that since the movement velocity between two joints was not fixed, phase portraits of CRP that correlated joint angles with velocities could be a better solution for assessing limb coordination. Therefore, within the protocol of CRP, our study aimed to quantify the upper-and lower-limb coordination of healthy preschoolers walking with curve-turning tasks (0°- 30°-60°), so as to comprehend: (1) preschoolers' coordination mechanisms during asymmetric motion; (2) their development of NMSS control in locomotion. According to literature, we hypothesized that preschoolers’ specific coordination mechanism that would differ from that of adults, particularly in the lower limbs, as they played a critical role in turning curves [3,9].

2.3. Data processing 2.3.1. Spatial-temporal parameters Spatial-temporal parameters (STP) were first evaluated. In each selected gait cycle, the initial heel strike and toe off were determined from the markers' trajectories on the heel and toe. Then, the stride length, stride time, stride cadence, Froude number and percentage of stance phase/swing phase were calculated. Stride length was measured according to the definition of Huxham et al. [15] and normalized by the individual's leg length according to Hof's theory [16]. The Froude number (Eq. (1)) was also used for a more precise comparison of stride velocity following the suggestion of Hof [16]. The mean and standard deviation were first obtained for participants and then within groups.

Froude Number =

v2 gl

(1)

where v represents walking velocity, g represents gravitational acceleration, and l represents leg length. 2.3.2. Continuous relative phase In terms of the CRP evaluation, time-series data were first filtered by a fourth-order low-pass Butterworth filter with a 6 Hz cut-off frequency; and a quintuple spline procedure [17] was used to create a 100 point time-normalized gait cycle (GC). Euler angles (θ) in a z-x-y coordinate sequence were computed and data in the sagittal plane of the ankle, knee, hip, elbow and thorax-humerus joint (THJ) (Table 1) were chosen for further analysis as suggested by literature [2,13,18,19]. In order to eliminate any artificial frequencies, we utilized Hilbert transform to calculate CRP, which was introduced by Lamb and stockl [20]. This method can be generally divided into four steps. (1) centering the Euler angle of each selected joint around zero (Eq. (2));

2. Method 2.1. Participants 45 healthy preschoolers and 10 healthy adults as a control group were recruited in this study. The mean age of preschoolers was 4.6 ± 1.1 y, the mean BMI was 16.0 ± 1.4, the mean leg length was 58.4 ± 2.5 cm, the gender ratio was 23 M / 22 F. The mean age of adults was 34.5 ± 9.8 y, the mean BMI was 20.9 ± 1.7, the mean leg length was 99.8 ± 7.3 cm, the gender ratio was 5 M/5 F. Criteria for selecting the participants were as follows: (1) no foot deformities or injuries; (2) the ability to walk independently and complete the whole test; (3) no abnormal gait patterns. This study was approved by the Ethics Committee of Sichuan University. All measures were executed after we disclosed study details to preschoolers' parents/adult participants and received their formal approval. Moreover, all the measurements and procedures followed the principles of the Helsinki Declaration.

θcentered (ti ) = θ (ti ) − min (θ (t )) −

max (θ (t )) − min (θ (t )) 2

(2)

where θcentered (ti ) and θ (ti ) are the zero-centered and original Euler angles at each time point (ti ) ; min (θ (t )) and max (θ (t )) are the minimum and maximum values over these 100 points, respectively. (2) transforming each zero-centered set of data into an analytic signal by Hilbert transform (Eq. (3));

ζ (t ) = θcentered (t ) + iH (t )

2.2. Kinematics measurement

(3)

where the Hilbert transform H (t ) of θcentered (t ) serves as the imaginary part of the analytic signal ζ (t ) . (3) calculating the phase angle (φ (ti ) ) at each point in time (Eq. (4));

The Coda Motion System (Coda Motion cx1, Charnwood Dynamics Ltd., United Kingdom) was used to obtain participants' kinematics data in both lower and upper limbs. Since we focused on the trailing leg rather than the supporting one, we designed a turning-left movement task. The left foot was considered the supporting leg and the motion in the trailing leg (right side) was studied. 23 key markers were set on the right side of the upper and lower limbs (Fig. 1 black point) and virtual points (Fig. 1 white point with black border) were computed via Odin software (V1.02, Charnwood Dynamics Ltd., United Kingdom). (More details of markers are shown in Appendix 1). Two collectors (Fig. 2) were aligned on the two sides of a six-meterlong walking track, at a 120-degree angle. Three curve-turning angles

φ (ti ) = tan−1 (

H (ti ) ) θcentered (ti )

(4)

(4) calculating the continuous relative phase between the two joints (Eq. (5)).

CRP(1 − 2) (ti ) = φ1 (ti ) − φ2 (ti )

(5)

where φ1 (ti ) represents the phase angle of the proximal joint and φ2 (ti ) represents the phase angle of the distal joint. A positive value indicates that the proximal joint leads the distal and vice versa [20]. According to the mathematical model above (Eqs. (2)–(5)), the CRP 2

Gait & Posture 75 (2020) 1–7

Q. Huang, et al.

Fig. 1. Demonstration of Markers’ Installation. The black points represent the 23 key markers set on the right side of the upper and lower limbs, the white points with black border represent virtual points computed according to Odin software.

Fig. 2. Sketch Map of Curving Turning. The definition of stride lengths (based on [17]) for 0 °C T and 60 °C T(30 °C T was the same)is shown. The solid line represents the edge of the walkway. The dashed lines (black for 0 °C T, dark gray for 30 °C T, light gray for 60 °C T) indicate the walking path assessed by markers. 3

Gait & Posture 75 (2020) 1–7

Q. Huang, et al.

Table 1 Definition of Each Euler Angle. Name

Spatial sequence

Definition

Ankle Knee Hip Elbow THJ

Z-X-Y Z-X-Y Z-X-Y Z-X-Y Z-X-Y

the the the the the

ankle angular values are the result of the foot movement relative to the shank. knee angular values are the result of the shank movement relative to the thigh. hip angular values are the result of the thigh movement relative to the pelvis. elbow angular values are the result of the forearm movement relative to the humerus. THJ angular values are the result of the humerus movement relative to the throax.

the majority variables in each curve-turning task. Preschoolers had a lower stride time, stride length and Froude number than adults; but their stride cadence was higher. Additionally, more stance phase and less swing phase were found for preschoolers. The two-way, mixed model ANOVA test revealed that the interaction of age and direction had no significant effect on all these variables (P > 0.05 for all).

between the knee (proximal segment) and ankle (distal segment) (CRP (Knee-Ankle)), hip and knee (CRP(Hip-Knee)) and THJ and elbow (CRP (THJ-Elbow) were quantified; then five trials were averaged upon each of the 100 normalized points of data. The CRP was stipulated as below: the range of distribution of the CRP falls within [-180°, 180°]; a value of 0° represents perfect in-phase behavior, which means the two joints are rotating in the same manner; the values of -180° and 180° both represent perfect anti-phase behavior, which means the two joints are rotating in an opposite manner [13,19,21]; Any angle between 0° and 180°/-180° indicates being out of phase [21]. Additionally, the variability of CRP (VCRP) was considered a valuable index to understand coordination and this index was used in this study to evaluate the variability within joints [1,22–24]. The VCRP was calculated as the between-stride standard deviation for a single subject within the 100 data points and then averaged across subjects [14,24]. A high value of VCRP indicates a more changeable degree of coordination between the two joints [25].

3.2. CRP outcomes As shown in Table 3, both age and direction, and their interaction revealed more significant effects on CRP values of Knee-Ankle coupling (%P = 83, %ES = 76 for age, %P = 91, %ES = 71 for direction and % P = 82, %ES = 70 for the age*direction interaction) and Hip-Knee coupling (%P = 55, %ES = 51 for age, %P = 68, %ES = 56 for direction and %P = 47, %ES = 46 for the age*direction interaction). While for THJ-Elbow coupling, apart from direction, these significant effects were lower (%P = 29, %ES = 14 for age, %P = 10, %ES = 4 for the age*direction interaction). According to Fig. 3, as the angle of curves increased, noticeable changes in CRP trajectories were found in the Knee-Ankle and Hip-Knee systems for preschoolers. Similar distributions were observed in adults. But in the THJ-Elbow coupling, preschoolers remained well in phase while turning curves; while adults displayed more out of phase motion in 10–45% and 60–100%GC. Inter-group variations (Table 3, results of independent T-test) further showed that %P between preschoolers and adults increased with curve angles in all coupling joints except for the Hip-Knee coupling when turning 60 degrees. The highest variations were found in the CRP (Knee-Ankle) and CRP (Hip-Knee) at each curve-turning task which occurred when the degree they were of out-phase was exaggerated or when the relative motion of joints was reversed. Nevertheless, basic patterns remained similar between preschoolers and adults. In terms of VCRP (Table 3), the two-way ANOVA test demonstrated that influences of age and age*direction on all coupling joints were low; but the direction had a higher effect on Knee-Ankle (%P = 74, % ES = 68) and Hip-Knee couplings (%P = 69, %ES = 57). Moreover, the mean VCRP values were found to increase with the intensification of curve angles for both groups; and those of Knee-Ankle and Hip-Knee couplings were higher in preschoolers than in adults. Negative findings were found in Upper-Limb-THJ-Elbow coupling. Moreover, group analysis revealed the inter-group differences remained at a relatively low level for both preschoolers and adults.

2.4. Statistical analysis Outcomes of one sample Kolmogorov-Smirnov Test showed that both STP variables and time series variables followed normal distribution; therefore, in terms of STP variables, a two-way analysis of variance (ANOVA) (interactions model: age, direction, and age*direction) was applied to explore the differences within turning tasks for preschoolers and adults. Meanwhile, in terms of time series variables, we chose GROUP ANALYSIS from the Model Statistic Procedure to look for variations between the two groups and within turning tasks in terms of CRP and VCRP from a macro perspective [26,27]. A two-way ANOVA test with LSD was also performed to assess the effects of age and direction on CRP and VCRP; while a T-test employing independent samples was used to explore detailed inter-group differences (preschoolers and adults) since it has a higher resolution for exploring variation. According to GROUP ANALYSIS, statistical analysis was executed on each set of time-series data, so we could generate 100 significant or insignificant outcomes. Then, we accumulated the percentage of points (%P) which were significantly different within turning tasks and between the two groups [26] to demonstrate the results of group analysis. Further, we applied effect size (ES) to assess the credibility of the group analysis [28]. The percentage of large ESs (%ES) (ES ＞0.8 for independent-samples T-test, ES＞0.5 for two-way ANOVA test) [28] was applied to determine meaningful differences for the variables of interest. All the statistical models were performed using SPSS (version 22, IBM, USA) with a significance level of 0.05 and a confidence interval of 95%.

4. Discussion In this study, we evaluated the gait and gait coordination of healthy preschoolers in curve-turning tasks from 0 degrees to 30 degrees and then to 60 degrees. During this turning process, we found preschoolers performed a specific gait pattern and applied certain coordination strategies; and this data was found to contrast with that of adults. By understanding those variations, we could gain insight into the NMSS development in the preschool population. At first, in order to deal with turning tasks, preschoolers modified their gait directly, such as increasing step frequency, shortening stride length and slowing walking velocity to maintain stability; meanwhile, they would take much more time in the stance phase in both the

3. Results 3.1. STP outcomes According to Table 2, as the curve angles increased (from 0° to 30° and 60 °CT), there were no significant differences within three CT tasks in stride time (P = 0.803), stride cadence(P = 0.714), stance phase or swing phase (P = 0.130); but CTs significantly decreased stride length and Froude number for both of the two groups (P = 0.000 for stride length, P = 0.000 for Froude number). Inter-group variations existed in 4

Gait & Posture 75 (2020) 1–7

Q. Huang, et al.

Table 2 Results of spatial-temporal parameters (STP) while curve turning (CT). Participant

Stride time (s) Stride cadence (step/min) Stride length Froude number Stance phase Swing phase

Direction (mean (SD))

Preschooler Adult Preschooler Adult Preschooler Adult Preschooler Adult Preschooler Adult Preschooler Adult

ANOVA

0 °C T

30 °C T

60 °C T

PAge

PDirection

PAge*Direction

0.96 (0.13) 1.10 (0.10) 125.64 (16.64) 109.70 (8.69) 1.30 (0.02) 1.34 (0.04) 0.22 (0.09) 0.26 (0.05) 59 (7) 52 (3) 41 (7) 48 (3)

0.97 (0.14) 1.10 (0.09) 127.28 (16.70) 109.72 (9.09) 0.83 (0.06) 0.87 (0.11) 0.09 (0.03) 0.11 (0.09) 60 (7) 54 (4) 40 (7) 46 (4)

0.93 (0.14) 1.10 (0.10) 131.98 (21.36) 110.07 (10.46) 0.78 (0.04) 0.81 (0.06) 0.09 (0.06) 0.10 (0.02) 62 (8) 53 (8) 38 (4) 47 (4)

0.000**

0.803

0.829

0.000**

0.714

0.767

0.210

0.000**

0.084

**

0.000

0.568

0.013*

0.130

0.187

0.013*

0.130

0.187

0.099

**

Analysis Results of Spatial-temporal parameters while FW and CT presented as follows: mean (standard deviation). * P value lower than 0.05. ** P value lower than 0.01. Table 3 Results of continuous relative phase (CRP) and variability of continuous relative phase (VCRP) while curve turning (CT). Coupling

Participant

Direction

ANOVA

0 °C T

CRP

Knee-Ankle Hip-Knee Elbow-THJ

VCRP

Knee-Ankle Hip-Knee Elbow-THJ

Preschooler Adult Preschooler Adult Preschooler Adult Preschooler Adult Preschooler Adult Preschooler Adult

30 °C T *

Mean

%P

%ES

37.95 42.07 −84.95 −88.79 44.29 43.04 11.48 10.74 10.7 10.23 9.58 9.38

76

65

38

26

16

8

28

14

16

5

11

3

**

60 °C T *

Mean

%P

%ES

51.85 58.22 −89.99 −99.57 46.64 48.79 14.02 12.37 11.47 11.12 9.33 10.55

87

81

56

55

44

31

47

38

21

7

25

13

**

Age *

**

Direction

%ES¶

71

82

70

68

56

47

46

19

53

31

10

4

48

37

74

68

31

28

27

30

21

69

57

17

12

21

22

16

33

30

16

15

%ES

¶

Mean

%P

%ES

%P

55.4 63.87 −97.03 −104.3 47.86 51.23 15.39 13.59 13.49 12.58 10.08 11.36

92

84

83

76

91

54

47

55

51

57

46

24

58

50

45 36

§

Age*Direction %P§

%P

§

%ES

¶

* percent of significant differences of CRP and VCRP between groups assessed by independent-samples T test. ** percent of large effect size (ES＞0.8) of CRP and VCRP between groups assessed by independent-samples T test. § percent of significant differences of CRP and VCRP of age and direction assessed by two-way ANOVA test. ¶ percent of large effect size(ES＞0.5) of CRP and VCRP assessed of age and direction assessed by two-way ANOVA test.

the trailing leg towards the direction of turning [3]. Meanwhile, the range of CRP of Knee-Ankle decreased in the stance phase which indicated the limitation of relative movement between Knee-Ankle. Constraining the Knee-Ankle coupling from the onset of the stance phase showed that preschoolers intended to take advantage of the forces generated by the stretch-shortening cycle of the muscles used in the flexion, storing more elastic energy for the propulsive phase [18,23]. However, the THJ-Elbow coupling maintained a constant relationship while turning curves. When contrasted with the adult group in each curve angle, a larger percentage of significant differences with moderate to high %ES values were obtained for preschoolers in KneeAnkle and Hip-Knee couplings and these percentages increased from 0 to 30 and then to 60 deg. We can assume that as the curve-turning angle increasing, children would activate more muscles involved in the hip and knee flexion, such as the gluteus maximus and gastrocnemius muscle, to generate enough force to rotate in the curving direction [23]. Thereby based on the outcomes of CRP, it was postulated that until age 6, in order to achieve the curve-turning tasks, preschoolers still performed specific gait strategies in key muscle utilization such as the timing of activation and muscle strength. Those also contributed to the differentiation in gait coordination between the preschoolers and adults. Additionally, the VCRP has been considered a valuable tool to

straight walking and the turning processes. All such efforts indicated that preschoolers usually require more stance time to generate enough force to turn the body, in order to accurately and successfully complete the task [6]. Our findings in the STP variables were also consistent with other relative studies [3,4,6,7]. Although this macro-analysis studied a general adaptive strategy of preschoolers in turning curves, detailed modifications within lower limbs and joints could not be explained. Then, CRP protocols were applied in this study to explore further variations in adaptive strategies of gait between preschoolers and adults. According to the results, we found that all CRPs were out-phase distributed. The CRPs in the coupling of the knee-ankle and elbow-THJ were in the positive area, while that of hip-knee in the negative area. It was implied that the knee/elbow (proximal segment) leads the ankle/ THJ (distal segment) while the hip (proximal segment) is behind the knee (distal segment). This confirmed that the knee is in a leading position while walking and it was consistent with current literature [19]. As the angle of the curve increased, in the lower limb of preschoolers, larger CRP values were found between the hip and knee (Table 3 and Fig. 3) implying that the hip flexion angle became exaggerated in the stance phase (more details of joint angles in Appendix 2). A higher flexion angle indicates that lower limbs help to generate enough counteracting centrifugal force and propulsive force to drive 5

Gait & Posture 75 (2020) 1–7

Q. Huang, et al.

Fig. 3. Portraits of Continuous Relative Phase (CRP) while curve turning (CT) for upper and lower limbs. A,C,E respectively indicate Knee-Ankle, Hip-Knee and THJ-Elbow couplings for preschooler group; B,D,F respectively indicate Knee-Ankle, Hip-Knee and THJ-Elbow couplings for adult group. The black solid line represents 0 °C T, gray solid line represents 30 °C T and black dashed line represents 60 °C T. For all the Portraits, toe-off was at 60%GC for both preschooler and adult groups.

6

Gait & Posture 75 (2020) 1–7

Q. Huang, et al.

understand coordination changes in locomotion [12], but others suggested that increased variability of coordination represents an unbalanced form of locomotion and a higher risk of injury [1,14,22,24]. VCRP was first found to increase alongside curve angle in all three couplings for both groups; and then a relatively large percentage of significant differences was found in the Knee-Ankle and (%P = 74, % ES = 68), Hip-Knee coupling (%P = 69%ES = 57). We think that the higher VCRP may be due to an altered visual condition alongside excessive joint motion and muscular activity during CT, which requires more effort for the neuromuscular system to regulate [1,3,23]. Moreover, the amplitude of adults was overall lower than that of preschoolers, and these findings suggest that preschoolers have a more variable relationship between the Knee-Ankle (%P = 48, %ES = 37) and Hip-Knee (%P = 30, %ES = 21) couplings. By contrast, preschoolers had a much lower VCRP in the Elbow-THJ coupling (% P = 22, %ES = 16), suggesting that preschoolers keep their upper limbs in a tense posture and limit their motion while turning curves. Therefore, according to the VCRP outcomes, we can infer that adults have better compensating strategies for direction change [29], while preschoolers are still in a gait fine-tuning period. Therefore, our hypothesis was proven. Limitations existed and our findings should be understood with caution. First, in our protocol, we only studied the trailing leg (right side). Since the conclusions from current research [3,6] indicate that the outer leg dominates curve-turning motion, our results were reliable. Secondly, the walking speed was not controlled and the participants walked and turned curves at their own preferred speed. Although speed would affect the frequencies in both lower and upper limbs, we applied a Hilbert transformation to filter out the effect of frequencies. Further research should be undertaken in the following areas: correlation and coordination between the trailing leg and the supporting one while turning curves.

[2]

[3] [4] [5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

[16]

5. Conclusions

[17]

Overall, this study demonstrated that in order to turn curves, preschoolers first chose to modify their STP, then to adjust coordination in Knee-Ankle, Hip-Knee, and THJ-Elbow coupled joints. Significant differences were still found in the STP and CRP variables between preschoolers and adults, and those variations implied that preschooler's ability to control their lower limbs is less sufficient and precise than that of adults. Therefore, we can postulate that preschoolers younger than 6 are still in the gait fine-tuning period and their NMSS mechanism is likewise in the development process.

[18]

[19]

[20]

[21]

Declaration of Competing Interest [22]

There were no conflicts of interest with other authors or institution for this study. [23]

Acknowledgements [24]

The authors appreciate all the children and their parents participated in this study; meanwhile we also thank the financial support from Natural Science Foundation of China (31700813) and China Postdoctoral Science Foundation (2015M571896).

[25]

[26]

Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.gaitpost.2019.09.013.

[27]

References

[28] [29]

[1] S. Mehdizadeh, A.R. Arshi, K. Davids, Quantifying coordination and coordination

7

variability in backward versus forward running: implications for control of motion, Gait Posture 42 (2015) 172–177, https://doi.org/10.1016/j.gaitpost.2015.05.006. S.L. Chiu, L.S. Chou, Effect of walking speed on inter-joint coordination differs between young and elderly adults, J. Biomech. 45 (2012) 275–280, https://doi.org/ 10.1016/j.jbiomech.2011.10.028. G. Courtine, M. Schieppati, Human walking along a curved path. I. Body trajectory, segment orientation and the effect of vision, Eur. J. Neurosci. 18 (2003) 177–190. G. Courtine, M. Schieppati, Human walking along a curved path. II. Gait features and EMG patterns, Eur. J. Neurosci. 18 (2003) 191–205. A. Ledebt, B. Bril, Y. Brenière, The build-up of anticipatory behaviourAn analysis of the development of gait initiation in children, Exp. Brain Res. 120 (1998) 9–17, https://doi.org/10.1007/s002210050372. P.C. Dixon, J. Stebbins, T. Theologis, A.B. Zavatsky, Spatio-temporal parameters and lower-limb kinematics of turning gait in typically developing children, Gait Posture 38 (2013) 870–875, https://doi.org/10.1016/j.gaitpost.2013.04.010. S.C. Strike, M.J.D. Taylor, The temporal–spatial and ground reaction impulses of turning gait: Is turning symmetrical? Gait Posture 29 (2009) 597–602, https://doi. org/10.1016/j.gaitpost.2008.12.015. C. Forsell, D. Conradsson, C. Paquette, E. Franzén, Reducing gait speed affects axial coordination of walking turns, Gait Posture 54 (2017) 71–75, https://doi.org/10. 1016/j.gaitpost.2017.02.020. A.E. Patla, A. Adkin, T. Ballard, Online steering: coordination and control of body center of mass, head and body reorientation, Exp. Brain Res. 129 (1999) 629–634, https://doi.org/10.1007/s002210050932. G. Courtine, M. Schieppati, Tuning of a basic coordination pattern constructs straight-ahead and curved walking in humans, J. Neurophysiol. 91 (2004) 1524–1535, https://doi.org/10.1152/jn.00817.2003. R. Grasso, P. Prévost, Y.P. Ivanenko, A. Berthoz, Eye-head coordination for the steering of locomotion in humans: an anticipatory synergy, Neurosci. Lett. 253 (1998) 115–118, https://doi.org/10.1016/S0304-3940(98)00625-9. J. Seay, J.M. Haddad, R.E.A. van Emmerik, J. Hamill, Coordination Variability Around the Walk to Run Transition During Human Locomotion, (2006), https://doi. org/10.1123/mcj.10.2.178. H.C. Yen, H.L. Chen, M.W. Liu, H.C. Liu, T.W. Lu, Age effects on the inter-joint coordination during obstacle-crossing, J. Biomech. 42 (2009) 2501–2506, https:// doi.org/10.1016/j.jbiomech.2009.07.015. J. Hamill, R.E.A. van Emmerik, B.C. Heiderscheit, L. Li, A dynamical systems approach to lower extremity running injuries, Clin. Biomech. 14 (1999) 297–308, https://doi.org/10.1016/S0268-0033(98)90092-4. F. Huxham, J. Gong, R. Baker, M. Morris, R. Iansek, Defining spatial parameters for non-linear walking, Gait Posture 23 (2006) 159–163, https://doi.org/10.1016/j. gaitpost.2005.01.001. A.L. Hof, Scaling gait data to body size, Gait Posture 4 (1996) 222–223, https://doi. org/10.1016/0966-6362(95)01057-2. J.E. Clark, S.J. Phillips, A longitudinal study of intralimb coordination in the first year of independent walking: a dynamical analysis, Child Dev. 64 (1993) 1143–1157. P. Dedieu, R. Chamoun, G. Lacaud, T. Moulinat, M. Queron, P.G. Zanone, Does a not-so-recent ankle sprain influence interjoint coordination during walking? Clin. Biomech. 50 (2017) 114–121, https://doi.org/10.1016/j.clinbiomech.2017.10.016. S. Ogaya, A. Iwata, Y. Higuchi, S. Fuchioka, The association between intersegmental coordination in the lower limb and gait speed in elderly females, Gait Posture 48 (2016) 1–5, https://doi.org/10.1016/j.gaitpost.2016.04.018. P.F. Lamb, M. Stöckl, On the use of continuous relative phase: review of current approaches and outline for a new standard, Clin. Biomech. 29 (2014) 484–493, https://doi.org/10.1016/j.clinbiomech.2014.03.008. J. Hamill, J.M. Haddad, W.J. McDermott, Issues in quantifying variability from a dynamical systems perspective, J. Appl. Biomech. 16 (2000) 407–418 http://www. embase.com/search/results?subaction=viewrecord&from=export&id= L32049418. T. Hein, T. Schmeltzpfenning, I. Krauss, C. Maiwald, T. Horstmann, S. Grau, Using the variability of continuous relative phase as a measure to discriminate between healthy and injured runners, Hum. Mov. Sci. 31 (2012) 683–694, https://doi.org/ 10.1016/j.humov.2011.07.008. M.J.R. Gittoes, W. Cassie, Intralimb joint coordination patterns of the lower extremity in maximal velocity phase sprint running, J. Appl. Biomech. 26 (2010) 188–195. R.H. Miller, S.A. Meardon, T.R. Derrick, J.C. Gillette, Continuous relative phase variability during an exhaustive run in runners with a history of iliotibial band syndrome, J. Appl. Biomech. 24 (2008) 262–270 http://www.ncbi.nlm.nih.gov/ pubmed/18843156. N. Stergiou, J.L. Jensen, B.T. Bates, S.D. Scholten, G. Tzetzis, A dynamical systems investigation of lower extremity coordination during running over obstacles, Clin. Biomech. 16 (2001) 213–221, https://doi.org/10.1016/S0268-0033(00)00090-5. J.D. Eggleston, M.R. Landers, B.T. Bates, E. Nagelhout, J.S. Dufek, Weighted walking influences lower extremity coordination in children on the autism Spectrum, Percept. Mot. Skills 125 (2018) 1103–1122, https://doi.org/10.1177/ 0031512518803178. J.S. Dufek, J.D. Eggleston, J.R. Harry, R.A. Hickman, A comparative evaluation of gait between children with autism and typically developing matched controls, Med. Sci. (Basel, Switzerland). 5 (2017) 1, https://doi.org/10.3390/medsci5010001. J. Cohen, Statistical Power ANALYSIS for the Behavioral Sciences, (1988), https:// doi.org/10.1016/B978-0-12-179060-8.50012-8. F. Lacquaniti, Y. Ivanenko, M. Zago, Kinematic Control of Walking, (2002).