- Email: [email protected]
Influence of treadmill acceleration on actual walk-to-run transition
Gait & Posture 31 (2010) 52–56
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Influence of treadmill acceleration on actual walk-to-run transition I. Van Caekenberghe a,*, V. Segers a,1, K. De Smet a,1, P. Aerts a,b,1,2, D. De Clercq a,3 a b
Ghent University, Department of Movement and Sports Sciences, Watersportlaan 2, B-9000 Gent, Belgium University of Antwerp, Campus Drie Eiken, Laboratory for Functional Morphology, Universiteitsplein 1, B-2610 Antwerpen, Belgium
A R T I C L E I N F O
A B S T R A C T
Article history: Received 23 January 2009 Received in revised form 24 August 2009 Accepted 31 August 2009
When accelerating continuously, humans spontaneously change from a walking to a running pattern by means of a walk-to-run transition (WRT). Results of previous studies indicate that when higher treadmill accelerations are imposed, higher WRT-speeds can be expected. By studying the kinematics of the WRT at different accelerations, the underlying mechanisms can be unravelled. 19 young, healthy female subjects performed walk-to-run transitions on a constantly accelerating treadmill (0.1, 0.2 and 0.5 m s2). A higher acceleration induced a higher WRT-speed, by effecting the preparation of transition, as well as the actual transition step. Increasing the acceleration caused a higher WRT-speed as a result of a greater step length during the transition step, which was mainly a consequence of a prolonged airborne phase. Besides this effect on the transition step, the direct preparation phase of transition (i.e. the last walking step before transition) appeared to fulfil specific constraints required to execute the transition regardless of the acceleration imposed. This highlights an important role for this step in the debate regarding possible determinants of WRT. In addition spatiotemporal and kinematical data confirmed that WRT remains a discontinuous change of gait pattern in all accelerations imposed. It is concluded that the walk-to-run transition is a discontinuous switch from walking to running which depends on the magnitude of treadmill belt acceleration. ß 2009 Elsevier B.V. All rights reserved.
Keywords: Gait transition Biomechanics Spatiotemporal Kinematics Acceleration Treadmill WRT Walk-to-run transition
1. Introduction Walking humans spontaneously switch from a walking to a running gait when a positive acceleration is imposed [1]. This walk-to-run transition (WRT) is frequently used as a paradigm to gain insights into the interaction of neural control, physical characteristics and intrinsic dynamics enabling human locomotion [2–4]. The mode of accelerating towards WRT is considered to affect WRT: a higher acceleration is believed to lead to a higher WRTspeed. This assumption originates from the comparison of WRT’s in and between studies that imposed accelerations. Nevertheless results of those studies are not equivocal. Several treadmill studies have compared the WRT using low constant accelerations, but were unable to reach a consensus whether WRT is influenced by acceleration. A higher WRT-speed
* Corresponding author. Tel.: +32 9 264 86 59; fax: +32 9 264 64 84. E-mail addresses: [email protected] (I. Van Caekenberghe), [email protected] (V. Segers), [email protected] (K. De Smet), [email protected] (P. Aerts), [email protected] (D. De Clercq). 1 Tel.: +32 9 264 63 12; fax: +32 9 264 64 84. 2 Tel.: +32 3 820 22 69; fax: +32 3 820 22 71. 3 Tel.: +32 9 264 63 22; fax: +32 9 264 64 84. 0966-6362/$ – see front matter ß 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.gaitpost.2009.08.244
with a higher magnitude of treadmill acceleration was indicated by Thorstensson and Roberthson [1] (WRT-speeds 1.89–1.92– 1.94 m s1, accelerations, respectively, 0.05–0.08–0.11 m s2) and Li [5] (WRT-speeds 2.25–2.28–2.32–2.34–2.40 m s1, accelerations 0.04–0.06–0.08–0.10–0.12 m s2). On the other hand, Segers et al. [6] did not observe an effect on WRT-speed (WRT-speeds 2.12–2.10– 2.16 m s1, accelerations 0.05–0.07–0.10 m s2). Other studies in which higher accelerations were imposed than those discussed above demonstrated a higher WRT-speed. De Smet et al. [7] investigated the spontaneous WRT over-ground in which subjects were able to accelerate in their own preferred manner. After gait initiation, subjects showed a constant acceleration around 0.5 m s2 in approach to transition. This preferred acceleration was substantially higher than any other acceleration ever investigated in a gait transition research design both on treadmill or over-ground (0.17 m s2) [8,9]. Compared to these studies at lower accelerations the preferred WRT-speed was higher (2.66 m s1) when using this higher acceleration. This difference in WRT-speed could also be caused by the over-ground condition. Nevertheless as acceleration is a crucial factor in the realization of the actual (i.e. using a ramped protocol) WRT, it is important to investigate its role in WRT. As in all studies on the actual transition from walking to running this transition manifests with a clear discontinuity in the
I. Van Caekenberghe et al. / Gait & Posture 31 (2010) 52–56
cyclic locomotion pattern, it is expected that the reason(s) for a raise in WRT-speed with a higher imposed acceleration can be detected in studying the kinematics of this discontinuity. Indeed, at constant accelerations ranging from 0.05 to 0.12 m s2 on treadmill [4,6,10,11] and 0.17 to 0.5 m s2 over-ground [7,12], WRT emerged as a discontinuity in locomotion, because the transition step is spatiotemporally, kinematically and dynamically [12] significantly different from the preceding walking and the following running step. During the transition step locomotion switches from a movement pattern showing a double stance phase, an extended knee of the stance leg at mid-stance and out-of-phase energy fluctuations of the centre of mass (COM) to a pattern exhibiting a flight phase (FP), a flexed knee of the stance leg at midstance and in-phase energy fluctuations of the COM. This paper aims to offer insights into following research questions. (1) Does the magnitude of treadmill acceleration influence the preferred WRT-speed? With a higher acceleration, a higher WRT-speed is expected. (2) How is the spatiotemporal (and kinematical) realization of the WRT affected by the magnitude of acceleration and can these mechanisms explain the increase in WRT-speed? In all accelerations a clear discontinuity is expected. 2. Methods The 19 subjects in this study were young and healthy females (body height = 170 cm (coefficient of variation, cv = 0.05), body weight = 64 kg (cv = 0.13), trochanter height = 90 cm (cv = 0.08), shoe size (EU) = 39.85 (cv = 0.04)), who were free from injury or disease at the time of participation. As such it is believed that, even though no tests were conducted, all subjects had normal strength and range of motion. All subjects were physically active, but not necessarily trained runners. Nine subjects indicated to run regularly, although training bouts were restricted to once a week or less. Two subjects indicated athletics as their main sport and practiced 6–8 h a week. They did not indicate whether this consisted of running or other events. All subjects signed informed consent forms. The ethical committee of the Ghent University Hospital approved of the experimental protocol. In the acceleration protocols treadmill belt speed increased with a constant acceleration of 0.1 m s2, 0.2 m s2 or 0.5 m s2 starting from the preferred walking speed until the preferred running speed (or higher if no transition occurred) was reached. Subjects were instructed to locomote on the treadmill in a comfortable way. Each acceleration condition consisted of five trials in which the same treadmill acceleration was imposed. Acceleration blocks were offered in a randomized order. Treadmill acceleration was validated using Maxtraq software by means of markers placed at equal distances on the belt. Slopes of linear regressions of treadmill velocity vs. time were not significantly different (one-sample t test) than the accelerations imposed. The effect of the treadmill acceleration on WRT was isolated from sources of bias. To minimize sound of the accelerating motor subjects wore headphones. To minimize potential fear being thrown off of the treadmill during high positive accelerations the treadmill was embedded in the floor. Prior to data collection subjects were familiarized to walking and running at different speeds and accelerating (varying from 0.1 to 0.5 m s2) on treadmill for at least 10 min. This time interval has proven to be sufficient for stabilization of sagittal plane kinematics and variability of spatiotemporal factors indicating adaptation to the treadmill [13,14]. For 19 subjects initial contact and toe-off were visually detected (using MaxTRAQ software) on sagittal video sequences recorded with two Bassler cameras (100 Hz) placed at treadmill height at approximately 1 m of the right side of the subject. One was focussed on the location of initial contact and the other on toe-off. Initial contact was defined as the first frame of foot-belt contact; toe-off as the last frame. Steps were numbered with respect to the transition step (step0, i.e. the first
53
step with a flight phase), walking steps prior to transition counting backwards (), running steps after the transition step counting forwards (+). For analyzing WRT as a discontinuity in gait, a comparison between the transition steps, the preceding step and the following step is necessary. Therefore, the spatiotemporal analyses start at initial contact of step1 and end at initial contact of step+2. Following procedures were followed in calculating spatiotemporal variables. (t = time, i = step number, a = imposed acceleration, VW = preferred walking speed, ic = initial contact, to = toe-off) Step frequency ðSFi Þ ¼
1 t ic iþ1 t ic i
Horizontal location of IC (Xic) was used as a means of correcting step length and subjects’ speed for movement in the laboratory reference frame. Step length ðSLi Þ ¼
mean treadmill speed during step i þ ðX ic iþ1 X ic i Þ SF
Speed at ic ðV ic i Þ ¼ treadmill speed at ic ðV ic i ¼ a t ic i þ V W Þ (As, given a certain Vic i, mean step speed (Vi) is higher for higher accelerations (i.e. the speed increment between tic and tic+1 is higher) this acceleration effect during the step is eliminated in the measure of instantaneous treadmill belt speed at ic) Mean step speed of the subject ðV i Þ ¼ SLi SFi WRT-speed ¼ mean speed of the subject at step0 ðV 0 Þ Flight phase duration ðFPÞ ¼ t ic iþ1 t to i Statistical analyses for spatiotemporal measures consisted of repeated measures (RM) ANOVAs using acceleration condition and step separately as a within factor. A comparison of main effects with Bonferroni-adjustment was made. Two-tailed pvalues were used except for the speed measures of step0. In order to detect the main contributor to speed a multiple linear regression of SL and SF (V = c SL + d SF, in which c and d represent the coefficients) was conducted separately for each acceleration condition in which a higher standardized beta-value points at more importance of this factor in the regression. For 13 of 19 subjects 3D-kinematics were recorded using an eight camera Qualisys Pro Reflex system (200 Hz). The following 46-marker design was used. Markers were bilaterally placed on 1st and 5th metatarsal, lateral and medial aspect of calcaneus, lateral and medial malleolus, anterior aspect of shank (2), lateral and medial femoral condyles, anterior aspect of thigh (2), greater trochanter, anterior superior iliac spine, sacral bone, C7, anterior aspect of cavitas glenoidalis, anterior aspect of upper arm (2), medial and lateral epicondyle of the humerus, fore arm (2) and the styloid process of the ulna. In order to calculate kinematics an 11-segment model (consisting of feet, shanks, thighs, trunk, upper arms and lower arms), which enabled to track the body COM, was developed using Visual3D. In both the kinematical and spatiotemporal analysis p-values were considered significant when p < 0.05 (**), a trend was observed when p < 0.10 (*).
3. Results When acceleration was higher a significantly higher WRTspeed was apparent when regarding both the mean step speed of step0 (Table 1) and the speed at initial contact of step0 (0.1 m s2 = 2.21 m s1, cv = 0.08; 0.2 m s2 = 2.25 m s1, cv = 0.09; 0.5 m s2 = 2.35 m s1, cv = 0.11; 0.1–0.5:**, 0.2–0.5:**) (effect sizes: 0.21– 1.00). This higher speed was paralleled by a larger step length and a higher step frequency of the transition step in the highest acceleration (p(accelerations) of step0, Table 2). Normalized bvalues of a multiple linear regression analysis were higher for step length than for step frequency, indicating that step length contributed more to the higher step speed (Table 3).
Table 1 Effect of acceleration on step speed. Mean (M) and coefficients of variation (CV) for V (mean speed of the subject) of steps 1, 0 and 1. Analysis: repeated measures (RM) ANOVA, comparison of main effects with Bonferroni-adjustment (1-tailed for step0, 2-tailed for steps 1 and +1). **: p < 0.05; *: p < 0.10. (n = 19).
V (m s1) Step1 Step0 Step+1
a = 0.1 m s2
a = 0.2 m s2
a = 0.5 m s2
M
M
M
2.07 2.09 2.32
CV
0.08 0.10 0.09
2.06 2.17 2.43
CV
0.10 0.12 0.09
2.07 2.31 2.78
CV
0.10 0.10 0.10
Pairwise comparisons RM
0.1 m s2 vs. 0.2 m s2
0.1 m s2 vs. 0.5 m s2
0.2 m s2 vs. 0.5 m s2
p
p
p
p
ns ** **
ns ** **
ns ** **
ns ** **
I. Van Caekenberghe et al. / Gait & Posture 31 (2010) 52–56
54
Table 2 Spatiotemporal variables of the realization of actual WRT means (M) and coefficients of variation (CV) of step frequency (SF) and step length (SL). Analyses (between steps and between accelerations): repeated measures (RM) ANOVA pairwise comparisons, comparison of main effects with Bonferroni-adjustment (two-tailed). **: p < 0.05; *: p < 0.10. (n = 19).
Table 3 Contribution of step length (SL) and step frequency (SF) to speed of the transition step (V0). Standardized beta coefficients (b) and coefficients of variance (R2) for SL and SF. Analysis: multiple linear regression. (n = 19).
SL (b standardized) SF (b standardized) R2
0.1 m s2
0.2 m s2
0.5 m s2
0.994 0.616 0.998
0.867 0.470 0.998
0.900 0.442 0.998
For a sub sample of 13 subjects step length of the transition step was broken down into five components (Fig. 1a), each relating to different phases in the step. These components are described in the caption of the figure and further elaborated in the second part of the addendum. With an increasing acceleration flight distance (2a, FLIGHT) and mean belt speed during the flight phase of step0 (Vtm flight 0) increased significantly. Flight phase duration of step0 (FP0) and vertical COM-velocity at toe-off of step0 (COMV vert to 0) were significantly lower for the lowest acceleration (Fig. 1b). For all accelerations step0 was spatiotemporally significantly different from the preceding walking and following running step (Table 2). In all accelerations step frequency was lower during the transition step as opposed to the preceding and following step. Step length was larger for the transition step as opposed to the preceding walking step. In step1 no significant effect of acceleration was found for the spatiotemporal variables investigated (Table 2). Kinematics of WRT in all accelerations and the accelerationrelated effects are presented in the addendum. 4. Discussion When a higher acceleration was imposed a significantly higher transition speed was found. This confirms results of Li [5] and expands them to an ecologically more valid range of accelerations [7]. The aim of this discussion is to highlight the mechanisms contributing to the higher transition speed. It will be demonstrated that causes can be found during and before the transition step.
4.1. During the transition step A higher acceleration leads to a higher transition speed. Although an effect of acceleration on step length and step frequency is only visible in the highest acceleration condition, the significantly different transition speed between the two lowest accelerations could be due to a combination of small nonsignificant increases in both variables. An increase in step length, rather than an increase in step frequency of the transition step is responsible for the higher transition speed with a higher acceleration. An increase in step length can theoretically be achieved by increasing at least one of the five parameters determining step length of the transition step on treadmill (Fig. 1a). Of those factors, a more pronounced flight distance (2a, FLIGHT) is proven to be responsible for the larger step length, since this distance is the only one which significantly increases with the rising acceleration (Fig. 1b). The flight distance is determined by the product of: (i) the belt speed during flight and (ii) the flight phase duration. Both factors increase: (i) belt speed during flight is significantly greater in a higher acceleration; (2) flight phase duration is significantly longer for the two highest accelerations. This prolonged airborne phase is caused by a higher upward COM-velocity at toe-off. The higher vertical velocity is probably obtained by pushing off more vigorously against the backward accelerating belt, which also results in a small forward movement in the lab-reference frame (DCOM). These acceleration effects on the spatiotemporal realization of step0 are accompanied by only subtle kinematical changes (as further elaborated in the addendum). Only at toe-off of the stance phase of stride0, thus the start of the flight phase of step0, a less pronounced ankle plantar flexion is observed when acceleration is higher. This small adaptation could probably induce a more vertical direction of the ground reaction force vector and as such be related to the increased flight distance (2a, FLIGHT) during step0. Since the other components of step length (Fig. 1b) are not influenced by the acceleration it is not surprising that kinematics during those phases are not influenced either. An enlarged flight phase thus seems to be crucial for the higher transition speed. When applying this theory to the results of Segers
I. Van Caekenberghe et al. / Gait & Posture 31 (2010) 52–56
55
Fig. 1. Kinematical strategies to influence step length of the transition step.(a). SL = 1 + 2a + 2b + 3 + Ft. Step length is equal to the distance between two successive contacts on the treadmill belt. This distance can be split up into five parts. In order to increase step length the subject could enlarge the distance: (1) between metatarsals and the body COM at toe-off of step0 (=COMME), (2) the belt travels during flight by (2a) prolonging the duration of the flight phase, and consequently the distance the belt travels if the subject were stationary in the lab-reference frame (=FLIGHT). This distance can additionally be increased by (2b) moving the body COM forward in the lab-reference frame during flight (=DCOM), (3) between the body COM and calcaneus at initial contact of step+1 (=COMCA), (Ft) is an anthropometric measure of the distance between calcaneus and metatarsals. It is thus constant and cannot be influenced to enlarge step length.Stick figures: (– –) leg1 at toe-off of step0, ( ) COM, (—) leg2 at heel contact of step+1, (*) COM.(b). Influence of acceleration. Means (M) and coefficients of variation (CV) of step length (SL) and components of step length of the transition step. The reported step length is the one calculated out of the spatiotemporal data (see formula in methods).The flight distance (2a = FLIGHT) is the product of: Vtm flight 0 = the mean treadmill belt speed during the flight phase of step0, and FP0 = duration of the flight phase of step0.The flight phase duration is dependent upon: COMV vert to 0 = the vertical velocity of the body centre of mass at toe-off of step0. Analysis: repeated measures (RM) ANOVA, comparison of main effects with Bonferroni-adjustment (two-tailed). **: p < 0.05; *: p < 0.10. (n=13).
et al. [6] additional evidence for this reasoning is found: the latter did not find an acceleration effect on flight phase duration, and as a result neither on the thereby determined transition speed. 4.2. Before the transition step Not only does acceleration influence the transition step itself, an effect of acceleration can also be observed in the step before transition, since the higher speed is already apparent (yet to a lesser extent) at the start of the transition step. The higher speed at initial contact of step0 may be related to the triggering mechanisms of WRT. It is argued that constraints for triggers to act, as well as for the launch of a new gait pattern, can only arise at a few discrete phases within a gait cycle, thus causing a latency between both events [15–18]. The acceleration-related speed increase between the instant of the actual trigger and the eventual launch of WRT would be larger in a higher acceleration, therefore inevitably resulting in a higher WRT-speed. Hreljac [16] stated that triggers for transition should, among other criteria, exhibit a critical value (i.e. a constant value under varying conditions). Considering this, it is remarkable that spatiotemporal parameters (speed, step length and step frequency), which might also influence possible triggers, are identical for all accelerations in step1. This perhaps points at specific constraints required to execute the transition step regardless of the acceleration imposed. A similar interpretation could be derived from the constant (acceleration independent) duration of the direct preparation of transition as discussed by Miller et al. [19]. Tuning this preparatory step1 is possibly of greater importance than the one of step0. Despite the acceleration-related spatiotemporal differences described, a key aspect of WRT remains unaffected: regardless of acceleration the transition step remains a clear spatiotemporal and kinematical (see addendum) discontinuity in gait.
(Ghent University) [6,7,9,10,20] we opted for a compatible female test population. In addition females usually have a lower body mass than males. As such they are less likely to induce changes in the imposed treadmill belt speed and acceleration. It is not clear whether different results could be expected in a male population. Preliminar analysis of non-published own data collected for a male population lead to similar conclusions as those drawn from the present female dataset, despite that absolute figures for transition speeds logically differ between both populations as these are coupled to the differences in morphometrics [21,22]. But as the acceleration influence on mechanics of the transition step or on the triggering mechanisms is not yet understood, no inferences about possible interactions between sex and acceleration can be made. 5. Conclusions As expected a higher treadmill acceleration leads to a higher WRT-speed, which is a consequence of effects during step1 and step0. In step1 speed is equal across all accelerations pointing at an important role for step1 which would be worthwhile investigating in function of the determinant question in WRT. In step0 a higher acceleration induces a larger step length. This larger step length is kinematically realized by extending the flight phase, in which the subject probably uses a more vigorous push-off which also aids in pushing off against the backward accelerating treadmill. Nevertheless its sensitivity to accelerations WRT remains in all accelerations a discontinuity in gait. Conflict of interest There is no conflict of interest.
4.3. Choice for a female population
Appendix A. Supplementary data
For reasons of homogeneity of the transition data thus far gathered at the laboratory for Movement and Sport Science
Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.gaitpost.2009.08.244.
56
I. Van Caekenberghe et al. / Gait & Posture 31 (2010) 52–56
References [1] Thorstensson A, Roberthson H. Adaptations to changing speed in human locomotion: speed of transition between walking and running. Acta Physiol Scan 1987;131:211–4. [2] Nishikawa K. Neuromechanics: an integrative approach for understanding motor control. Integr Comp Biol 2007;47:16. [3] Farley CT, Ferris DP. Biomechanics of walking and running: center of mass movements to muscle action. Exerc Sport Sci Rev 1998;26:253–85. [4] Diedrich FJ, Warren JWH. Why change gaits? Dynamics of the walk-run transition. J Exp Psychol Hum 1995;21:183–203. [5] Li L. Stability landscapes of walking and running near gait transition speed. J Appl Biomech 2000;16:428–35. [6] Segers V, Aerts P, Lenoir M, De Clercq D. Spatiotemporal characteristics of the walk-to-run and run-to-walk transition when gradually changing speed. Gait Posture 2006;24:247–54. [7] De Smet K, Segers V, Lenoir M, De Clercq D. Spatiotemporal characteristics of spontaneous walk-to-run transition. Gait Posture 2009;29:54–8. [8] Hreljac A, Imamura R, Escamilla RF, Edwards WB. Effects of changing protocol, grade, and direction on the preferred gait transition speed during human locomotion. Gait Posture 2007;25:419–24. [9] Segers V, Aerts P, Lenoir M, De Clercq D. External forces during actual acceleration across transition speed. J Appl Biomech 2008;24:340–50. [10] Segers V, Lenoir M, Aerts P, De Clercq D. Kinematics of the transition between walking and running when gradually changing speed. Gait Posture 2007;26:349–61. [11] Hreljac A, Imamura RT, Escamilla RF, Edwards WB. When does a gait transition occur during human locomotion? J Sport Sci Med 2007;6:36–43.
[12] Segers V, Aerts P, Lenoir M, De Clerq D. Dynamics of the body centre of mass during actual acceleration across transition speed. J Exp Biol 2007;210:578– 85. [13] Matsas A, Taylor N, McBurney H. Knee joint kinematics from familiarised treadmill walking can be generalised to overground walking in young unimpaired subjects. Gait Posture 2000;11:46–53. [14] Lavcanska V, Taylor NF, Schache AG. Familiarization to treadmill running in young unimpaired adults. Hum Mov Sci 2005;24:544–57. [15] Neptune RR, Sasaki K. Ankle plantar flexor force production is an important determinant of the preferred walk-to-run transition speed. J Exp Biol 2005;208:799–809. [16] Hreljac A. Determinants of the gait transition speed during human locomotion: kinematic factors. J Biomech 1995;28:669–77. [17] Hreljac A, Imamura RT, Escamilla RF, Edwards WB, MacLeod T. The relationship between joint kinetic factors and the walk-run gait transition speed during human locomotion. J Appl Biomech 2008;24:149–57. [18] Malcolm P, Segers V, Van Caekenberghe I, De Clercq D. Experimental study of the influence of the m. tibialis anterior on the walk-to-run transition by means of a powered ankle-foot exoskeleton. Gait Posture 2008;29:6–10. [19] Miller J, Li L, Evans N. Gait transition speeds with different acceleration. J Sport Exercise Psychol 2000;22:S56–7. [20] Segers V, Aerts P, Lenoir M, De Clercq D. Dynamics of the body centre of mass during actual acceleration across transition speed. J Exp Biol 2007;210:578– 85. [21] Hreljac A. Effects of physical characteristics on the gait transition speed during human locomotion. Hum Mov Sci 1995;14:205–16. [22] Getchell N, Whitall J. Transitions to and from asymmetrical gait patterns. J Mot Behav 2004;36:13–27.
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Influence of treadmill acceleration on actual walk-to-run transition I. Van Caekenberghe a,*, V. Segers a,1, K. De Smet a,1, P. Aerts a,b,1,2, D. De Clercq a,3 a b
Ghent University, Department of Movement and Sports Sciences, Watersportlaan 2, B-9000 Gent, Belgium University of Antwerp, Campus Drie Eiken, Laboratory for Functional Morphology, Universiteitsplein 1, B-2610 Antwerpen, Belgium
A R T I C L E I N F O
A B S T R A C T
Article history: Received 23 January 2009 Received in revised form 24 August 2009 Accepted 31 August 2009
When accelerating continuously, humans spontaneously change from a walking to a running pattern by means of a walk-to-run transition (WRT). Results of previous studies indicate that when higher treadmill accelerations are imposed, higher WRT-speeds can be expected. By studying the kinematics of the WRT at different accelerations, the underlying mechanisms can be unravelled. 19 young, healthy female subjects performed walk-to-run transitions on a constantly accelerating treadmill (0.1, 0.2 and 0.5 m s2). A higher acceleration induced a higher WRT-speed, by effecting the preparation of transition, as well as the actual transition step. Increasing the acceleration caused a higher WRT-speed as a result of a greater step length during the transition step, which was mainly a consequence of a prolonged airborne phase. Besides this effect on the transition step, the direct preparation phase of transition (i.e. the last walking step before transition) appeared to fulfil specific constraints required to execute the transition regardless of the acceleration imposed. This highlights an important role for this step in the debate regarding possible determinants of WRT. In addition spatiotemporal and kinematical data confirmed that WRT remains a discontinuous change of gait pattern in all accelerations imposed. It is concluded that the walk-to-run transition is a discontinuous switch from walking to running which depends on the magnitude of treadmill belt acceleration. ß 2009 Elsevier B.V. All rights reserved.
Keywords: Gait transition Biomechanics Spatiotemporal Kinematics Acceleration Treadmill WRT Walk-to-run transition
1. Introduction Walking humans spontaneously switch from a walking to a running gait when a positive acceleration is imposed [1]. This walk-to-run transition (WRT) is frequently used as a paradigm to gain insights into the interaction of neural control, physical characteristics and intrinsic dynamics enabling human locomotion [2–4]. The mode of accelerating towards WRT is considered to affect WRT: a higher acceleration is believed to lead to a higher WRTspeed. This assumption originates from the comparison of WRT’s in and between studies that imposed accelerations. Nevertheless results of those studies are not equivocal. Several treadmill studies have compared the WRT using low constant accelerations, but were unable to reach a consensus whether WRT is influenced by acceleration. A higher WRT-speed
* Corresponding author. Tel.: +32 9 264 86 59; fax: +32 9 264 64 84. E-mail addresses: [email protected] (I. Van Caekenberghe), [email protected] (V. Segers), [email protected] (K. De Smet), [email protected] (P. Aerts), [email protected] (D. De Clercq). 1 Tel.: +32 9 264 63 12; fax: +32 9 264 64 84. 2 Tel.: +32 3 820 22 69; fax: +32 3 820 22 71. 3 Tel.: +32 9 264 63 22; fax: +32 9 264 64 84. 0966-6362/$ – see front matter ß 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.gaitpost.2009.08.244
with a higher magnitude of treadmill acceleration was indicated by Thorstensson and Roberthson [1] (WRT-speeds 1.89–1.92– 1.94 m s1, accelerations, respectively, 0.05–0.08–0.11 m s2) and Li [5] (WRT-speeds 2.25–2.28–2.32–2.34–2.40 m s1, accelerations 0.04–0.06–0.08–0.10–0.12 m s2). On the other hand, Segers et al. [6] did not observe an effect on WRT-speed (WRT-speeds 2.12–2.10– 2.16 m s1, accelerations 0.05–0.07–0.10 m s2). Other studies in which higher accelerations were imposed than those discussed above demonstrated a higher WRT-speed. De Smet et al. [7] investigated the spontaneous WRT over-ground in which subjects were able to accelerate in their own preferred manner. After gait initiation, subjects showed a constant acceleration around 0.5 m s2 in approach to transition. This preferred acceleration was substantially higher than any other acceleration ever investigated in a gait transition research design both on treadmill or over-ground (0.17 m s2) [8,9]. Compared to these studies at lower accelerations the preferred WRT-speed was higher (2.66 m s1) when using this higher acceleration. This difference in WRT-speed could also be caused by the over-ground condition. Nevertheless as acceleration is a crucial factor in the realization of the actual (i.e. using a ramped protocol) WRT, it is important to investigate its role in WRT. As in all studies on the actual transition from walking to running this transition manifests with a clear discontinuity in the
I. Van Caekenberghe et al. / Gait & Posture 31 (2010) 52–56
cyclic locomotion pattern, it is expected that the reason(s) for a raise in WRT-speed with a higher imposed acceleration can be detected in studying the kinematics of this discontinuity. Indeed, at constant accelerations ranging from 0.05 to 0.12 m s2 on treadmill [4,6,10,11] and 0.17 to 0.5 m s2 over-ground [7,12], WRT emerged as a discontinuity in locomotion, because the transition step is spatiotemporally, kinematically and dynamically [12] significantly different from the preceding walking and the following running step. During the transition step locomotion switches from a movement pattern showing a double stance phase, an extended knee of the stance leg at mid-stance and out-of-phase energy fluctuations of the centre of mass (COM) to a pattern exhibiting a flight phase (FP), a flexed knee of the stance leg at midstance and in-phase energy fluctuations of the COM. This paper aims to offer insights into following research questions. (1) Does the magnitude of treadmill acceleration influence the preferred WRT-speed? With a higher acceleration, a higher WRT-speed is expected. (2) How is the spatiotemporal (and kinematical) realization of the WRT affected by the magnitude of acceleration and can these mechanisms explain the increase in WRT-speed? In all accelerations a clear discontinuity is expected. 2. Methods The 19 subjects in this study were young and healthy females (body height = 170 cm (coefficient of variation, cv = 0.05), body weight = 64 kg (cv = 0.13), trochanter height = 90 cm (cv = 0.08), shoe size (EU) = 39.85 (cv = 0.04)), who were free from injury or disease at the time of participation. As such it is believed that, even though no tests were conducted, all subjects had normal strength and range of motion. All subjects were physically active, but not necessarily trained runners. Nine subjects indicated to run regularly, although training bouts were restricted to once a week or less. Two subjects indicated athletics as their main sport and practiced 6–8 h a week. They did not indicate whether this consisted of running or other events. All subjects signed informed consent forms. The ethical committee of the Ghent University Hospital approved of the experimental protocol. In the acceleration protocols treadmill belt speed increased with a constant acceleration of 0.1 m s2, 0.2 m s2 or 0.5 m s2 starting from the preferred walking speed until the preferred running speed (or higher if no transition occurred) was reached. Subjects were instructed to locomote on the treadmill in a comfortable way. Each acceleration condition consisted of five trials in which the same treadmill acceleration was imposed. Acceleration blocks were offered in a randomized order. Treadmill acceleration was validated using Maxtraq software by means of markers placed at equal distances on the belt. Slopes of linear regressions of treadmill velocity vs. time were not significantly different (one-sample t test) than the accelerations imposed. The effect of the treadmill acceleration on WRT was isolated from sources of bias. To minimize sound of the accelerating motor subjects wore headphones. To minimize potential fear being thrown off of the treadmill during high positive accelerations the treadmill was embedded in the floor. Prior to data collection subjects were familiarized to walking and running at different speeds and accelerating (varying from 0.1 to 0.5 m s2) on treadmill for at least 10 min. This time interval has proven to be sufficient for stabilization of sagittal plane kinematics and variability of spatiotemporal factors indicating adaptation to the treadmill [13,14]. For 19 subjects initial contact and toe-off were visually detected (using MaxTRAQ software) on sagittal video sequences recorded with two Bassler cameras (100 Hz) placed at treadmill height at approximately 1 m of the right side of the subject. One was focussed on the location of initial contact and the other on toe-off. Initial contact was defined as the first frame of foot-belt contact; toe-off as the last frame. Steps were numbered with respect to the transition step (step0, i.e. the first
53
step with a flight phase), walking steps prior to transition counting backwards (), running steps after the transition step counting forwards (+). For analyzing WRT as a discontinuity in gait, a comparison between the transition steps, the preceding step and the following step is necessary. Therefore, the spatiotemporal analyses start at initial contact of step1 and end at initial contact of step+2. Following procedures were followed in calculating spatiotemporal variables. (t = time, i = step number, a = imposed acceleration, VW = preferred walking speed, ic = initial contact, to = toe-off) Step frequency ðSFi Þ ¼
1 t ic iþ1 t ic i
Horizontal location of IC (Xic) was used as a means of correcting step length and subjects’ speed for movement in the laboratory reference frame. Step length ðSLi Þ ¼
mean treadmill speed during step i þ ðX ic iþ1 X ic i Þ SF
Speed at ic ðV ic i Þ ¼ treadmill speed at ic ðV ic i ¼ a t ic i þ V W Þ (As, given a certain Vic i, mean step speed (Vi) is higher for higher accelerations (i.e. the speed increment between tic and tic+1 is higher) this acceleration effect during the step is eliminated in the measure of instantaneous treadmill belt speed at ic) Mean step speed of the subject ðV i Þ ¼ SLi SFi WRT-speed ¼ mean speed of the subject at step0 ðV 0 Þ Flight phase duration ðFPÞ ¼ t ic iþ1 t to i Statistical analyses for spatiotemporal measures consisted of repeated measures (RM) ANOVAs using acceleration condition and step separately as a within factor. A comparison of main effects with Bonferroni-adjustment was made. Two-tailed pvalues were used except for the speed measures of step0. In order to detect the main contributor to speed a multiple linear regression of SL and SF (V = c SL + d SF, in which c and d represent the coefficients) was conducted separately for each acceleration condition in which a higher standardized beta-value points at more importance of this factor in the regression. For 13 of 19 subjects 3D-kinematics were recorded using an eight camera Qualisys Pro Reflex system (200 Hz). The following 46-marker design was used. Markers were bilaterally placed on 1st and 5th metatarsal, lateral and medial aspect of calcaneus, lateral and medial malleolus, anterior aspect of shank (2), lateral and medial femoral condyles, anterior aspect of thigh (2), greater trochanter, anterior superior iliac spine, sacral bone, C7, anterior aspect of cavitas glenoidalis, anterior aspect of upper arm (2), medial and lateral epicondyle of the humerus, fore arm (2) and the styloid process of the ulna. In order to calculate kinematics an 11-segment model (consisting of feet, shanks, thighs, trunk, upper arms and lower arms), which enabled to track the body COM, was developed using Visual3D. In both the kinematical and spatiotemporal analysis p-values were considered significant when p < 0.05 (**), a trend was observed when p < 0.10 (*).
3. Results When acceleration was higher a significantly higher WRTspeed was apparent when regarding both the mean step speed of step0 (Table 1) and the speed at initial contact of step0 (0.1 m s2 = 2.21 m s1, cv = 0.08; 0.2 m s2 = 2.25 m s1, cv = 0.09; 0.5 m s2 = 2.35 m s1, cv = 0.11; 0.1–0.5:**, 0.2–0.5:**) (effect sizes: 0.21– 1.00). This higher speed was paralleled by a larger step length and a higher step frequency of the transition step in the highest acceleration (p(accelerations) of step0, Table 2). Normalized bvalues of a multiple linear regression analysis were higher for step length than for step frequency, indicating that step length contributed more to the higher step speed (Table 3).
Table 1 Effect of acceleration on step speed. Mean (M) and coefficients of variation (CV) for V (mean speed of the subject) of steps 1, 0 and 1. Analysis: repeated measures (RM) ANOVA, comparison of main effects with Bonferroni-adjustment (1-tailed for step0, 2-tailed for steps 1 and +1). **: p < 0.05; *: p < 0.10. (n = 19).
V (m s1) Step1 Step0 Step+1
a = 0.1 m s2
a = 0.2 m s2
a = 0.5 m s2
M
M
M
2.07 2.09 2.32
CV
0.08 0.10 0.09
2.06 2.17 2.43
CV
0.10 0.12 0.09
2.07 2.31 2.78
CV
0.10 0.10 0.10
Pairwise comparisons RM
0.1 m s2 vs. 0.2 m s2
0.1 m s2 vs. 0.5 m s2
0.2 m s2 vs. 0.5 m s2
p
p
p
p
ns ** **
ns ** **
ns ** **
ns ** **
I. Van Caekenberghe et al. / Gait & Posture 31 (2010) 52–56
54
Table 2 Spatiotemporal variables of the realization of actual WRT means (M) and coefficients of variation (CV) of step frequency (SF) and step length (SL). Analyses (between steps and between accelerations): repeated measures (RM) ANOVA pairwise comparisons, comparison of main effects with Bonferroni-adjustment (two-tailed). **: p < 0.05; *: p < 0.10. (n = 19).
Table 3 Contribution of step length (SL) and step frequency (SF) to speed of the transition step (V0). Standardized beta coefficients (b) and coefficients of variance (R2) for SL and SF. Analysis: multiple linear regression. (n = 19).
SL (b standardized) SF (b standardized) R2
0.1 m s2
0.2 m s2
0.5 m s2
0.994 0.616 0.998
0.867 0.470 0.998
0.900 0.442 0.998
For a sub sample of 13 subjects step length of the transition step was broken down into five components (Fig. 1a), each relating to different phases in the step. These components are described in the caption of the figure and further elaborated in the second part of the addendum. With an increasing acceleration flight distance (2a, FLIGHT) and mean belt speed during the flight phase of step0 (Vtm flight 0) increased significantly. Flight phase duration of step0 (FP0) and vertical COM-velocity at toe-off of step0 (COMV vert to 0) were significantly lower for the lowest acceleration (Fig. 1b). For all accelerations step0 was spatiotemporally significantly different from the preceding walking and following running step (Table 2). In all accelerations step frequency was lower during the transition step as opposed to the preceding and following step. Step length was larger for the transition step as opposed to the preceding walking step. In step1 no significant effect of acceleration was found for the spatiotemporal variables investigated (Table 2). Kinematics of WRT in all accelerations and the accelerationrelated effects are presented in the addendum. 4. Discussion When a higher acceleration was imposed a significantly higher transition speed was found. This confirms results of Li [5] and expands them to an ecologically more valid range of accelerations [7]. The aim of this discussion is to highlight the mechanisms contributing to the higher transition speed. It will be demonstrated that causes can be found during and before the transition step.
4.1. During the transition step A higher acceleration leads to a higher transition speed. Although an effect of acceleration on step length and step frequency is only visible in the highest acceleration condition, the significantly different transition speed between the two lowest accelerations could be due to a combination of small nonsignificant increases in both variables. An increase in step length, rather than an increase in step frequency of the transition step is responsible for the higher transition speed with a higher acceleration. An increase in step length can theoretically be achieved by increasing at least one of the five parameters determining step length of the transition step on treadmill (Fig. 1a). Of those factors, a more pronounced flight distance (2a, FLIGHT) is proven to be responsible for the larger step length, since this distance is the only one which significantly increases with the rising acceleration (Fig. 1b). The flight distance is determined by the product of: (i) the belt speed during flight and (ii) the flight phase duration. Both factors increase: (i) belt speed during flight is significantly greater in a higher acceleration; (2) flight phase duration is significantly longer for the two highest accelerations. This prolonged airborne phase is caused by a higher upward COM-velocity at toe-off. The higher vertical velocity is probably obtained by pushing off more vigorously against the backward accelerating belt, which also results in a small forward movement in the lab-reference frame (DCOM). These acceleration effects on the spatiotemporal realization of step0 are accompanied by only subtle kinematical changes (as further elaborated in the addendum). Only at toe-off of the stance phase of stride0, thus the start of the flight phase of step0, a less pronounced ankle plantar flexion is observed when acceleration is higher. This small adaptation could probably induce a more vertical direction of the ground reaction force vector and as such be related to the increased flight distance (2a, FLIGHT) during step0. Since the other components of step length (Fig. 1b) are not influenced by the acceleration it is not surprising that kinematics during those phases are not influenced either. An enlarged flight phase thus seems to be crucial for the higher transition speed. When applying this theory to the results of Segers
I. Van Caekenberghe et al. / Gait & Posture 31 (2010) 52–56
55
Fig. 1. Kinematical strategies to influence step length of the transition step.(a). SL = 1 + 2a + 2b + 3 + Ft. Step length is equal to the distance between two successive contacts on the treadmill belt. This distance can be split up into five parts. In order to increase step length the subject could enlarge the distance: (1) between metatarsals and the body COM at toe-off of step0 (=COMME), (2) the belt travels during flight by (2a) prolonging the duration of the flight phase, and consequently the distance the belt travels if the subject were stationary in the lab-reference frame (=FLIGHT). This distance can additionally be increased by (2b) moving the body COM forward in the lab-reference frame during flight (=DCOM), (3) between the body COM and calcaneus at initial contact of step+1 (=COMCA), (Ft) is an anthropometric measure of the distance between calcaneus and metatarsals. It is thus constant and cannot be influenced to enlarge step length.Stick figures: (– –) leg1 at toe-off of step0, ( ) COM, (—) leg2 at heel contact of step+1, (*) COM.(b). Influence of acceleration. Means (M) and coefficients of variation (CV) of step length (SL) and components of step length of the transition step. The reported step length is the one calculated out of the spatiotemporal data (see formula in methods).The flight distance (2a = FLIGHT) is the product of: Vtm flight 0 = the mean treadmill belt speed during the flight phase of step0, and FP0 = duration of the flight phase of step0.The flight phase duration is dependent upon: COMV vert to 0 = the vertical velocity of the body centre of mass at toe-off of step0. Analysis: repeated measures (RM) ANOVA, comparison of main effects with Bonferroni-adjustment (two-tailed). **: p < 0.05; *: p < 0.10. (n=13).
et al. [6] additional evidence for this reasoning is found: the latter did not find an acceleration effect on flight phase duration, and as a result neither on the thereby determined transition speed. 4.2. Before the transition step Not only does acceleration influence the transition step itself, an effect of acceleration can also be observed in the step before transition, since the higher speed is already apparent (yet to a lesser extent) at the start of the transition step. The higher speed at initial contact of step0 may be related to the triggering mechanisms of WRT. It is argued that constraints for triggers to act, as well as for the launch of a new gait pattern, can only arise at a few discrete phases within a gait cycle, thus causing a latency between both events [15–18]. The acceleration-related speed increase between the instant of the actual trigger and the eventual launch of WRT would be larger in a higher acceleration, therefore inevitably resulting in a higher WRT-speed. Hreljac [16] stated that triggers for transition should, among other criteria, exhibit a critical value (i.e. a constant value under varying conditions). Considering this, it is remarkable that spatiotemporal parameters (speed, step length and step frequency), which might also influence possible triggers, are identical for all accelerations in step1. This perhaps points at specific constraints required to execute the transition step regardless of the acceleration imposed. A similar interpretation could be derived from the constant (acceleration independent) duration of the direct preparation of transition as discussed by Miller et al. [19]. Tuning this preparatory step1 is possibly of greater importance than the one of step0. Despite the acceleration-related spatiotemporal differences described, a key aspect of WRT remains unaffected: regardless of acceleration the transition step remains a clear spatiotemporal and kinematical (see addendum) discontinuity in gait.
(Ghent University) [6,7,9,10,20] we opted for a compatible female test population. In addition females usually have a lower body mass than males. As such they are less likely to induce changes in the imposed treadmill belt speed and acceleration. It is not clear whether different results could be expected in a male population. Preliminar analysis of non-published own data collected for a male population lead to similar conclusions as those drawn from the present female dataset, despite that absolute figures for transition speeds logically differ between both populations as these are coupled to the differences in morphometrics [21,22]. But as the acceleration influence on mechanics of the transition step or on the triggering mechanisms is not yet understood, no inferences about possible interactions between sex and acceleration can be made. 5. Conclusions As expected a higher treadmill acceleration leads to a higher WRT-speed, which is a consequence of effects during step1 and step0. In step1 speed is equal across all accelerations pointing at an important role for step1 which would be worthwhile investigating in function of the determinant question in WRT. In step0 a higher acceleration induces a larger step length. This larger step length is kinematically realized by extending the flight phase, in which the subject probably uses a more vigorous push-off which also aids in pushing off against the backward accelerating treadmill. Nevertheless its sensitivity to accelerations WRT remains in all accelerations a discontinuity in gait. Conflict of interest There is no conflict of interest.
4.3. Choice for a female population
Appendix A. Supplementary data
For reasons of homogeneity of the transition data thus far gathered at the laboratory for Movement and Sport Science
Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.gaitpost.2009.08.244.
56
I. Van Caekenberghe et al. / Gait & Posture 31 (2010) 52–56
References [1] Thorstensson A, Roberthson H. Adaptations to changing speed in human locomotion: speed of transition between walking and running. Acta Physiol Scan 1987;131:211–4. [2] Nishikawa K. Neuromechanics: an integrative approach for understanding motor control. Integr Comp Biol 2007;47:16. [3] Farley CT, Ferris DP. Biomechanics of walking and running: center of mass movements to muscle action. Exerc Sport Sci Rev 1998;26:253–85. [4] Diedrich FJ, Warren JWH. Why change gaits? Dynamics of the walk-run transition. J Exp Psychol Hum 1995;21:183–203. [5] Li L. Stability landscapes of walking and running near gait transition speed. J Appl Biomech 2000;16:428–35. [6] Segers V, Aerts P, Lenoir M, De Clercq D. Spatiotemporal characteristics of the walk-to-run and run-to-walk transition when gradually changing speed. Gait Posture 2006;24:247–54. [7] De Smet K, Segers V, Lenoir M, De Clercq D. Spatiotemporal characteristics of spontaneous walk-to-run transition. Gait Posture 2009;29:54–8. [8] Hreljac A, Imamura R, Escamilla RF, Edwards WB. Effects of changing protocol, grade, and direction on the preferred gait transition speed during human locomotion. Gait Posture 2007;25:419–24. [9] Segers V, Aerts P, Lenoir M, De Clercq D. External forces during actual acceleration across transition speed. J Appl Biomech 2008;24:340–50. [10] Segers V, Lenoir M, Aerts P, De Clercq D. Kinematics of the transition between walking and running when gradually changing speed. Gait Posture 2007;26:349–61. [11] Hreljac A, Imamura RT, Escamilla RF, Edwards WB. When does a gait transition occur during human locomotion? J Sport Sci Med 2007;6:36–43.
[12] Segers V, Aerts P, Lenoir M, De Clerq D. Dynamics of the body centre of mass during actual acceleration across transition speed. J Exp Biol 2007;210:578– 85. [13] Matsas A, Taylor N, McBurney H. Knee joint kinematics from familiarised treadmill walking can be generalised to overground walking in young unimpaired subjects. Gait Posture 2000;11:46–53. [14] Lavcanska V, Taylor NF, Schache AG. Familiarization to treadmill running in young unimpaired adults. Hum Mov Sci 2005;24:544–57. [15] Neptune RR, Sasaki K. Ankle plantar flexor force production is an important determinant of the preferred walk-to-run transition speed. J Exp Biol 2005;208:799–809. [16] Hreljac A. Determinants of the gait transition speed during human locomotion: kinematic factors. J Biomech 1995;28:669–77. [17] Hreljac A, Imamura RT, Escamilla RF, Edwards WB, MacLeod T. The relationship between joint kinetic factors and the walk-run gait transition speed during human locomotion. J Appl Biomech 2008;24:149–57. [18] Malcolm P, Segers V, Van Caekenberghe I, De Clercq D. Experimental study of the influence of the m. tibialis anterior on the walk-to-run transition by means of a powered ankle-foot exoskeleton. Gait Posture 2008;29:6–10. [19] Miller J, Li L, Evans N. Gait transition speeds with different acceleration. J Sport Exercise Psychol 2000;22:S56–7. [20] Segers V, Aerts P, Lenoir M, De Clercq D. Dynamics of the body centre of mass during actual acceleration across transition speed. J Exp Biol 2007;210:578– 85. [21] Hreljac A. Effects of physical characteristics on the gait transition speed during human locomotion. Hum Mov Sci 1995;14:205–16. [22] Getchell N, Whitall J. Transitions to and from asymmetrical gait patterns. J Mot Behav 2004;36:13–27.