Journal of Electromyography and Kinesiology 25 (2015) 709–714

Contents lists available at ScienceDirect

Journal of Electromyography and Kinesiology

journal homepage: www.elsevier .com/locate / je lek in

Effects of body weight unloading on electromyographic activity duringoverground walking

http://dx.doi.org/10.1016/j.jelekin.2015.05.0011050-6411/� 2015 Elsevier Ltd. All rights reserved.

⇑ Corresponding author at: Biorobotics and Biomechanics Lab, Department ofMechanical Engineering, Technion-Israel Institute of Technology, Haifa 32000,Israel. Tel.: +972 54 4991287; fax: +972 4 8295187, +972 4 8295711.

E-mail address: ariellef@technion.ac.il (A.G. Fischer).

Arielle G. Fischer ⇑, Eytan M. Debbi, Alon WolfBiorobotics and Biomechanics Lab, Department of Mechanical Engineering, Technion-Israel Institute of Technology, Haifa, Israel

a r t i c l e i n f o a b s t r a c t

Article history:Received 4 March 2015Received in revised form 19 April 2015Accepted 8 May 2015

Keywords:Body weight unloadingElectromyographyBiomechanicsGait analysisRehabilitation

Background: Body weight unloading (BWU) on treadmills is a common method of gait rehabilitation.However, treadmills slightly but significantly modify gait biomechanical parameters thus confound theeffects of BWU. By conducting our experiments under conditions that replicate daily walking and control-ling for speed variability, with a mechanical device designed to pull the BWU system at a constant speed,this study could assess the unique effects of BWU on gait electromyography (EMG) of healthy subjects.Methods: Fifteen healthy subjects walked overground in a control (no suspension vest) and three (0%,15%, 30%) BWU experimental conditions. The EMG activity of the Tibialis Anterior (TA), LateralGastrocnemius (LG), Vastus Lateralis (VL), and Rectus Femoris (RF) were recorded (six trials per condi-tion). Results: ANOVA showed significant differences in the peak activity and integrated EMG of the TA,LG and VL. Pairwise comparisons of EMG parameters under 0% vs. 15% and 15% vs. 30% BWU levelsshowed that the increase in BWU levels decreased the peak and integrated EMG of the TA, LG, and VLwithout pattern modification. Conclusion: Overground gait with up to 30% BWU reduces joint loads with-out modifying the muscle activation patterns. Several clinical applications for overground gait reeduca-tion with BWU are suggested.

� 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Body weight reduction with a body weight unloading (BWU)suspension vest during treadmill walking is a common method ofgait rehabilitation for patients with neurological and muscu-loskeletal impairments (Barbeau et al., 2004; Hesse et al., 1994;Lee and Hidler, 2008; Mangione et al., 1996; Patiño et al., 2007;Threlkeld et al., 2003). The assumption behind this method is thatby alleviating the load on lower joints and the related pain experi-enced, partial body weight unloading will motivate clinical sub-jects to undergo gait rehabilitation to induce sensory stimulation(Threlkeld et al., 2003), strengthen muscles and improve gaitspeed, balance and locomotion right after injury or surgery(Dickstein, 2008; Lamontagne and Fung, 2004; Perry and Davids,1992; Perry et al., 1995; Schmid et al., 2007; Sousa et al., 2009;Van Hedel et al., 2006).

Past research assumed that treadmill and overground gaitbiomechanical parameters were similar enough so that gait correc-tions on treadmills would easily transfer to daily overground

walking (Arsenault et al., 1986; Lee and Hidler, 2008; Murrayet al., 1985; Riley et al., 2007; van Ingen Schenau, 1979).However, research refutes this assumption by showing significantdifferences in biomechanical parameters in these walking modali-ties (Alton et al., 1998; Arsenault et al., 1986; Lee and Hidler, 2008;Murray et al., 1985). Differences in muscle activation patterns wereobserved in the Quadriceps (Alton et al., 1998; Arsenault et al.,1986; Lee and Hidler, 2008; Murray et al., 1985), the TibialisAnterior, and the Hamstrings, (Lee and Hidler, 2008; Riley et al.,2007; Savelberg et al., 1998). These findings led researchers to con-clude that the unique effects of BWU on gait parameters could notbe assessed as long as the walking modality (overground vs. tread-mill) was not controlled. The recommendation was to conduct gaitresearch with BWU overground under conditions that replicatedaily walking.

One of the main obstacles in the study on the effects of BWUduring overground gait was healthy subjects’ inability to maintaina comfortable gait speed when having to pull the BWU system towhich they were suspended (Patiño et al., 2007). To control forsubjects’ speed variability a specially tailored mechanical devicewas designed in our laboratory to pull the BWU system at a con-stant speed, thereby allow healthy subjects to maintain a comfort-able speed under various levels of BWU. Once the gait modality

710 A.G. Fischer et al. / Journal of Electromyography and Kinesiology 25 (2015) 709–714

and speed variability were controlled, this study examined theunique effects of partial BWU on healthy subjects’ lower limb mus-cle activation patterns during overground walking.

We hypothesized that the reduction of healthy subjects’ bodyweight by 0%, 15% and 30% during overground walking at a con-stant speed will (1) decrease the EMG parameters i.e., peak EMGactivity and area under the curve (integrated EMG) of the lowerjoints muscles over the gait cycle and (2) EMG activation patternsof lower joints muscles will not be affected.

2. Methods

2.1. Subjects

Fifteen healthy male subjects were recruited for this study.Selection criteria were being healthy with no previous orthopedic,musculoskeletal or neurological pathology, and having similaranthropometric profile (i.e. weight, shoe size, height, dominantleg). Subjects mean age in years was 23.5 ± 3 (S.D.), their meanheight was 1.73 ± 0.06 m and the mean weight was 67.9 ± 9.5 kg.This study was approved by the Institutional Review Board, andinformed consent was obtained from all subjects prior to datacollection.

2.2. Instrumentation

The Biodex Unweighing System (Biodex Co., Shirley, NY)(Fig. 1a) was used to manipulate subjects’ body weight and accom-modate for the vertical displacement of the center of gravity duringgait under controlled BWU levels. This system includes a suspen-sion vest with shoulder straps, a pelvic belt and a groin pieceattached to the belt. Once in an upright position in the BWU device,

Fig. 1. (a) The BWU Biodex system. (b) The BWU

a pulley lifted the subjects until the predetermined BWU level wasreached and indicated the amount of weight reduced (in kg) on thescreen of the Biodex system.

Past research assessing the effectiveness of BWU on joint mus-cle strengthening have shown that a body weight reduction above30% BWU resulted in distortions of the EMG activation patternswith a significant decrease in EMG amplitudes of the Quadriceps,the Calves and the Hamstrings muscles activity during stance(Colby et al., 1999; Lee and Hidler, 2008). Also observed were mod-ifications in EMG activity duration (i.e. changes in EMG on–off tim-ing) of the Tibialis Anterior and Gastrocnemius which significantlyincreased (Finch et al., 1991). Learning from previous research(Patiño et al., 2007) up to 30% body weight reduction was appliedin this research.

A potentially confounding variable interacting with the effectsof BWU was gait speed. Research conducted on treadmills withBWU, during which healthy subjects’ gait speed could easily becontrolled, showed that a speed lower than 0.7 m/s resulted in sig-nificant changes in muscle EMG patterns as exhibited by theincrease in co-contraction of flexor and extensor muscles(Gastrocnemius and Quadriceps) and the longer activation of flexormuscles (Hamstrings and Tibialis Anterior) (Van Hedel et al., 2006).A direct relationship was also observed between gait speedand EMG activity and magnitude of the soleus, and theGastrocnemius muscles, with an increase in speed resulting in anincrease of EMG activity of these muscles (Lewek, 2011). Gaitspeed was however hard to control overground, given that healthysubjects were unable to maintain a comfortable walking speedwhen pulling the BWU system to which they were suspended(Patiño et al., 2007). To circumvent this problem, we designed anew mechanical device composed of an electric winch installedin the wall facing the Biodex system. A cable, connecting the winchvia a pulley system to the Biodex Unweighing Device, pulled the

system connected to the electrical winch.

A.G. Fischer et al. / Journal of Electromyography and Kinesiology 25 (2015) 709–714 711

system across the floor set at a constant speed (Fig. 1b). This speedwas set to 1.1–1.2 m/s since it corresponded to the average over-ground walking speed of the research participants. The validity ofthis measure was corroborated in previous research findings(Bohannon, 1997). In an accommodation period, a metronomehelped subjects maintain a comfortable gait speed that matchedthe speed at which the winch pulled BWU system.

2.3. Procedure

The walking task included a control (no vest) and three experi-mental conditions, 0%, 15% and 30% BWU, manipulated by havingsubjects wear the Biodex suspension vest, and a control conditionwith no vest. Subjects were instructed to walk overground along a10 m walkway while stepping on two AMTI OR6-7-1000 forceplates placed in tandem along a walkway under the experimentaland control conditions. Each one of the control and BWU experi-mental conditions included six trials.

2.4. Electromyographic data measurement and parameters

An eight-camera infrared Vicon motion analysis system withVicon Nexus software (Oxford Metrics Ltd., UK), included two forceplates, was used to simultaneously collect and transfer the EMGand force plate data to the computer. Recording of the lowerextremities EMG activity was performed by placing two AurionZeroWire wireless surface EMG electrodes (Aurion, Milan, Italy)on the belly of four muscles. The surface EMG electrodes werepositioned on the following muscles: Tibialis Anterior (TA),Lateral Gastrocnemius (LG), Vastus Lateralis (VL), and RectusFemoris (RF) (Fig. 2). After cleaning the skin with isopropyl alcoholthe electrodes were placed according to the recommendations ofSENIAM (surface EMG for non-invasive assessment of muscles)on the four muscles.

The EMG measures included the duration of EMG activity, thepeak EMG activity, and the integrated EMG signal for each of the

Fig. 2. Placement of the EMG electrodes on a subject’s leg: (a) lateral view, (b)anterior view.

muscles examined: TA, LG, VL, and RF under the control and exper-imental conditions of 0%, 15% and 30% BWU.

2.5. Data analysis

Data from the force plates were analyzed to determine the ini-tial heel strike and toe off, and divided into gait cycles accordingly.Each gait cycle corresponded to the interval between two consec-utive contacts of the ipsilateral heel with the force plate. The EMGvalues were averaged across six trials to obtain one mean value pervariable for each subject under the control and three 0%, 15% and30% BWU experimental conditions. EMG data were subsequentlynormalized to the full gait cycle such that 0% corresponded tothe touch of the ipsilateral heel on the contact surface and 100%to the subsequent touch by the same heel.

Muscle activity analysis (Konrad, 2005; Nigg et al., 2006; Shiaviet al., 1998) included the calculation of the linear envelopes of theEMG signals computed by using an algorithm that included succes-sive operations. To do so the resting baseline activity was sub-tracted from the active walking trials. Afterwards the signal wasfiltered with a 4th order band-pass filter (BPF), with cutoff frequen-cies of 10 and 500 Hz. This signal was then rectified, before beingsmoothed with a low-pass filter of 50 ms. The dependent EMGparameters for each one of the muscles under study (Section 2.4)were calculated by using a custom developed Matlab (MATLAB,2012) algorithm. The duration of the EMG for the following mus-cles: TA, VL, RF was determined operationally as starting fromthe beginning of the EMG signal activation (at heel strike) up untilthe signal went below 5% of the maximum EMG signal. For the GCmuscles the onset and offset were hand selected starting when thesignal went above 5% of the max value until it went below 5% maxvalue. The parameters peak and integrated EMG were examinedwithin the duration interval of the signal. The maximum value ofthe EMG and integrated EMG for each muscle was calculated whenthe subject walked in the control condition, with no harness (NH).The values of the maximum EMG and area under the curveobtained for each one of the experimental 0%, 15% and 30% BWUconditions were expressed as a fraction of the values obtainedfor these parameters under the no-harness control condition(Romkes et al., 2006).

2.6. Statistical analysis

The parameters were analyzed for statistical differences usingrepeated measures analysis of variance (ANOVA) conducted withall the dependent measures with the three experimental BWU con-ditions as independent measures with subsequent post-hoc t-testsperformed for pair-wise comparison. Given that the values of theexperimental conditions were expressed as a fraction of valuesobtained in the control condition, these values were not includedin the analyses. The analyses were performed using Matlab soft-ware. The correlation between muscle activation trajectories com-pared pairwise under the control and three experimental BWUconditions (0%, 15% and 30%) was assessed with the Pearson’s cor-relation coefficient (R) and root mean square errors (RMSE).

3. Results

3.1. Tibialis anterior

ANOVA (2 df) conducted on the EMG parameters of the TAunder three experimental conditions 0%, 15% and 30% BWU indi-cated a significant (p < 0.05) effect of conditions on the integratedEMG and peak EMG activity of the TA. T-test paired comparisons ofthe experimental conditions 0% vs. 15% BWU and 15% vs. 30% BWU

Table 2Mean values (±SD) of the Lateral Gastrocnemius (LG) under 0%, 15% and 30% BWUconditions (N = 15).

Experimental conditions

0% 15% 30%

EMG parameters of LGArea under curve 153.49 ± 113.4 140.53 ± 113.9 b 121.37 ± 105.8Peak activity 0.7 ± 0.3a 0.57 ± 0.3b 0.43 ± 0.2Duration 48.47 ± 10.9 50.48 ± 11.7 48.11 ± 10.7

a Significant difference (p < 0.05) between 0% and 15% BWU.b Significant difference (p < 0.05) between 15% and 30% BWU.

712 A.G. Fischer et al. / Journal of Electromyography and Kinesiology 25 (2015) 709–714

showed no significant differences in the duration of EMG activity ofthe TA. In contrast, a significant (p < 0.05) decrease in the peakactivity and integrated EMG of the TA was indicated under 0% vs.15% BWU and 15% vs. 30% BWU (Table 1 and Fig. 3).

3.2. Lateral gastrocnemius

ANOVA (2 df) conducted on EMG parameters of LG under threeexperimental conditions 0%, 15% and 30% BWU indicated a signif-icant (p < 0.05) effect of conditions on the dependent measuresthe integrated EMG and peak EMG activity. T-test paired compar-isons of the experimental conditions showed significant reductions(p < 0.02) in the peak activity of the LG when the experimental con-ditions 0% vs. 15% BWU were compared. Similarly, t-test compar-isons of LG parameters under 15% vs. 30% BWU conditionsindicated a significant decrease in the integrated EMG (p < 0.01)and peak activity (p < 0.005). No significant difference in durationof LG activity was indicated (Table 2 and Fig. 4).

3.3. Vastus lateralis

ANOVA (2 df) conducted on EMG parameters of VL under threeexperimental conditions 0% 15% and 30% BWU indicated signifi-cant differences between conditions on the dependent variablesintegrated EMG and peak EMG activity. T-tests pairwise compar-isons of the EMG parameters of the VL under the experimental0% vs. 15% and 15% vs. 30% BWU conditions showed no significantdifferences in the duration of EMG activity. In contrast, the peakactivity and integrated EMG of the VL were found to significantlydecrease under 0% vs. 15% BWU (p < 0.001). T-test comparisonsof these parameters under 15% vs. 30% BWU also showed a

Table 1Mean values (±SD) of the Tibialis Anterior (TA) under 0%, 15% and 30% BWUconditions (N = 15).

Experimental conditions

0% 15% 30%

EMG parameters of the TAArea under curve 102.62 ± 64a 78.5 ± 40b 51.92 ± 23Peak activity 1.13 ± 0.4a 0.96 ± 0.4b 0.75 ± 0.3Duration 18.8 ± 6.8 17.72 ± 8.8 16.44 ± 5.9

a Significant difference (p < 0.05) between 0% and 15% BWU.b Significant difference (p < 0.05) between 15% and 30% BWU.

Fig. 3. EMG signal of the Tibialis Anterior (TA) under 0%, 15% and 30% BWUconditions.

significant decrease for the integrated EMG of the VL (p < 0.01)(Table 3 and Fig. 5).

3.4. Rectus femoris

ANOVA (2 df) was conducted on EMG parameters of the RFshowed no effect of conditions (0%, 15%, 30% BWU) on the inte-grated EMG, peak activity and duration of EMG activity of thismuscle (Fig. 6).

3.5. Muscle activation trajectories

Table 4 shows highly significant (p < 0.001) correlations(0.84 < R < 0.92) for all EMG trajectory patterns of the TA, LG, VL,and RF compared pairwise under the experimental conditions 0%vs. 15% BWU, and 15% vs. 30% BWU. Additionally, the RMSEobtained were small ranging from 0.07 to 0.12 (Fig. 3–6).

Fig. 4. EMG signal of the Lateral Gastrocnemius (LG) under 0%, 15% and 30% BWUconditions.

Table 3Mean values (±SD) of the Vastus Lateralis (VL) under 0%, 15% and 30% BWU conditions(N = 15).

Experimental conditions

0% 15% 30%

EMG parameters of VLArea under curve 85.26 ± 46.28 a 62.73 ± 38.21b 35.66 ± 29.2Peak activity 0.65 ± 0.22a 0.54 ± 0.21 0.42 ± 0.29Duration (% GC) 21.7 ± 5.18 19.11 ± 6.04 17.87 ± 6.65

a Significant difference (p < 0.05) between 0% and 15% BWU.b Significant difference (p < 0.05) between 15% and 30% BWU.

Fig. 5. EMG signal of the Vastus Lateralis (VL) under 0%, 15% and 30% BWUconditions.

Fig. 6. EMG signal of the Rectus Femoris (RF) under 0%, 15% and 30% BWUconditions.

Table 4Pairwise correlations of the lower extremities muscle patterns of the followingmuscles: TA, LG, VL and RF under 0% vs. 15% BWU conditions, and 15% vs. 30% BWUconditions.

0–15% BWU 15–30% BWUMeasures

TA r 0.87* 0.91*

RMSE 0.12 0.11LG r 0.9* 0.84*

RMSE 0.07 0.08VL r 0.92* 0.9*

RMSE 0.07 0.08RF r 0.86* 0.89*

RMSE 0.12 0.12

* p < 0.001; root mean square errors RMSE.

A.G. Fischer et al. / Journal of Electromyography and Kinesiology 25 (2015) 709–714 713

4. Discussion

This study examined the unique effects of 0%, 15% and 30% BWUon the EMG activation patterns and timing of the lower joint mus-cles of healthy subjects walking overground at a comfortablespeed. Research comparing healthy subjects’ gait parameters on

treadmill vs. overground concluded that treadmill walking doesnot replicate daily overground walking given the small but never-theless significant changes observed in biomechanical parametersand EMG activity in these two modalities. Healthy subjects walk-ing on a treadmill exhibited a significant decrease of the BicepsFemoris and Tibialis Anterior EMG activity throughout stance.Hamstrings, Vastus medialis and Adductor Longus were alsoobserved to decrease during the swing phase (Arsenault et al.,1986; Lee and Hidler, 2008; Murray et al., 1985). In comparisonto overground, EMG gait parameters on treadmills exhibited achange in the timing of EMG burst of the Hamstrings occurring ear-lier at onset of the stance phase (Lee and Hidler, 2008; Murrayet al., 1985). This effect was especially pronounced at extremelylow gait speeds, i.e. under 0.7 m/s (Nymark et al., 2005).Consequently, only by conducting gait studies overground, underconditions that replicate daily walking, could one assess the uniqueeffects of partial BWU on gait EMG parameters. This conclusionwas supported by several other researchers (Lewek, 2011;Threlkeld et al., 2003; Van Hedel et al., 2006). Healthy subjects’speed variability during overground walking with a BWU vestwas another confounding variable that was controlled prior to con-ducted this study. A specially tailored mechanical device, thatpulled the BWU system at a constant speed, enabled the subjectsto maintain a comfortable overground speed even when suspendedby a BWU vest.

The first hypothesis of this study assumed that overgroundwalking with up to 30% BWU will modify the EMG parameters-integrated EMG and peak activity of the lower limb muscles. Thishypothesis was confirmed for the muscles TA, LG and VL with a sig-nificant inverse relationship indicated between an increase in BWUlevels and a decrease in the peak EMG activity and integrated EMGvalues of these muscles. Similar findings in EMG activity of the TAand LG were reported by Colby et al. (1999) and by Van Hedel et al.(2006). The findings of this study also showed that for the weightaccepting flexors and extensors muscles, TA and LG muscles, themost pronounced modification in the peak EMG activity and inte-grated EMG occurred at the loading portion of the gait cycle(stance phase). In contrast the duration of the EMG of the musclesexamined remained unchanged during overground walking withup to 30% BWU level.

The second hypothesis of this study was also confirmed. Areduction of up to 30% BWU did not change the EMG signal pat-terns over a gait cycle of the leg muscles of healthy subjects walk-ing overground at a comfortable speed. The high correlationbetween EMG signal trajectories compared pairwise under theexperimental conditions suggest that a reduction of up to 30% ofhealthy subjects’ body weight significantly reduces the EMGparameters of TA, LG and VL parameters without modifying thepatterns of the EMG signal trajectories.

This study has broad implications in the field of gait rehabilita-tion since it suggests that clinical subjects suffering from muscleweakness may safely train with a reduction of up to 30% of theirbody weight. Colby et al. (1999) noted that a decrease inGastrocnemius muscle activity amplitude under increasing BWUlevels may be essential at push off, when ankle plantarflexion isactivated to propel the body forward. At the stance phase such adecrease may also facilitate the proper phasing of this muscle forclinical patients undergoing rehabilitation. Similarly, the decreasein amplitude of the TA under increasing BWU may be of impor-tance given the difficulties in dorsiflexion exhibited by clinicalpatients. Once patients’ muscles are strengthened in the courseof overground in situ rehabilitation with BWU and achieve greaterbalance and locomotion activity, BWU may be gradually reduceduntil patients resume daily walking without BWU.

One of the main limitations of this study relates to the limitednature of the sample, which included only healthy young male

714 A.G. Fischer et al. / Journal of Electromyography and Kinesiology 25 (2015) 709–714

adults and therefore the results are characteristic of that testedgroup. Further studies expanding our sample to other populationwill increase the reliability of our findings.

Future research may chart the modifications and changes inhealthy subjects’ lower limb EMG during the gait cycle duringoverground walking under 0%, 15% and 30% BWU. This chartingwill allow to produce a normative data set that could be used todetermine which muscles of the lower limbs are most affectedby BWU and what are the deviations of clinical subjects’ EMGparameters and the gait corrections that need to be achieved.

Conflict of interest

The authors declare that they have no conflict of interest

Acknowledgements

The authors thank Brenda Geiger, Ph.D. and Ariel Dowling, Ph.D.for their technical assistance and constructive comments.

References

Alton F, Baldey L, Caplan S, Morrissey MC. A kinematic comparison of overgroundand treadmill walking. Clin Biomech 1998;13(6):434–40.

Arsenault AB, Winter DA, Marteniuk RG. Treadmill versus walkway locomotion inhumans: an EMG study. Ergonomics 1986;29(5):665–76.

Barbeau H, Lamontagne A, Ladouceur M, Mercier I, Fung J. Optimizing locomotorfunction with body weight support training and functional electricalstimulation. Progress in motor control. Champaign (IL): Human Kinetics;2004. p. 237–51.

Bohannon RW. Comfortable and maximum walking speed of adults aged 20–79years: reference values and determinants. Age Ageing 1997;26(1):15–9.

Colby SM, Kirkendall DT, Bruzga RF. Electromyographic analysis and energyexpenditure of harness supported treadmill walking: implications for kneerehabilitation. Gait Posture 1999;10(3):200–5.

Dickstein R. Rehabilitation of gait speed after stroke: a critical review ofintervention approaches. Neurorehab Neural Repair 2008;22(6):649–60.

Finch L, Barbeau H, Arsenault B. Influence of body weight support on normal humangait: development of a gait retraining strategy. Phys Therapy1991;71(11):842–55.

Hesse S, Bertelt C, Schaffrin A, Malezic M, Mauritz KH. Restoration of gait innonambulatory hemiparetic patients by treadmill training with partial body-weight support. Arch Phys Med Rehabilit 1994;75(10):1087–93.

Konrad P. The abc of emg. A practical introduction to kinesiologicalelectromyography; 2005. p. 1.

Lamontagne A, Fung J. Faster is better implications for speed-intensive gait trainingafter stroke. Stroke 2004;35(11):2543–8.

Lee SJ, Hidler J. Biomechanics of overground vs. treadmill walking in healthyindividuals. J Appl Phys 2008;104(3):747–55.

Lewek MD. The influence of body weight support on ankle mechanics duringtreadmill walking. J Biomech 2011;44(1):128–33.

Mangione KK, Axen K, Haas F. Mechanical unweighting effects on treadmill exerciseand pain in elderly people with osteoarthritis of the knee. Phys Therapy1996;76(4):387–94.

MATLAB, 2012. version 7.14 (R2012a). Natick, Massachusetts: The MathWorks Inc.Murray M, Spurr G, Sepic S, Gardner G, Mollinger L. Treadmill vs. floor walking:

kinematics, electromyogram, and heart rate. J Appl Physiol 1985;59(1):87–91.Nigg B, Hintzen S, Ferber R. Effect of an unstable shoe construction on lower

extremity gait characteristics. Clin Biomech 2006;21(1):82–8.Nymark JR, Balmer SJ, Melis EH, Lemaire ED, Millar S. Electromyographic and

kinematic nondisabled gait differences at extremely slow overground andtreadmill walking speeds. J Rehabilit Res Dev 2005;42(4):523.

Patiño M, Gonçalves A, Monteiro B, Santos I, Barela A, Barela J. Kinematic, kineticand electromyographic characteristics of young adults walking with andwithout partial body weight support. Revista Brasileira Fisioterapia2007;11(1):19–25.

Perry J, Davids JR. Gait analysis: normal and pathological function. J PediatricOrthopaedics 1992;12(6):815.

Perry J, Garrett M, Gronley JK, Mulroy SJ. Classification of walking handicap in thestroke population. Stroke 1995;26(6):982–9.

Riley PO, Paolini G, Della Croce U, Paylo KW, Kerrigan DC. A kinematic and kineticcomparison of overground and treadmill walking in healthy subjects. GaitPosture 2007;26(1):17–24.

Romkes J, Rudmann C, Brunner R. Changes in gait and EMG when walking with theMasai Barefoot Technique. Clin Biomech (Bristol, Avon) 2006;21(1):75–81.

Savelberg HH, Vorstenbosch MA, Kamman EH, van de Weijer, Jolanda GW,Schambardt HC. Intra-stride belt-speed variation affects treadmill locomotion.Gait Posture 1998;7(1):26–34.

Schmid A, Duncan PW, Studenski S, Lai SM, Richards L, Perera S, et al. Improvementsin speed-based gait classifications are meaningful. Stroke2007;38(7):2096–100.

Shiavi R, Frigo C, Pedotti A. Electromyographic signals during gait: criteria forenvelope filtering and number of strides. Med Biol Eng Comput1998;36(2):171–8.

Sousa C, Barela J, Prado-Medeiros C, Salvini T, Barela A. The use of body weightsupport on ground level: an alternative strategy for gait training of individualswith stroke. J Neuroeng Rehabilit 2009;6(1):43.

Threlkeld AJ, Cooper LD, Monger BP, Craven AN, Haupt HG. Temporospatial andkinematic gait alterations during treadmill walking with body weightsuspension. Gait Posture 2003;17(3):235–45.

Van Hedel H, Tomatis L, Müller R. Modulation of leg muscle activity and gaitkinematics by walking speed and bodyweight unloading. Gait Posture2006;24(1):35–45.

van Ingen Schenau G. Some fundamental aspects of the biomechanics of overgroundversus treadmill locomotion. Med Sci Sports Exercise 1979;12(4):257–61.

Arielle Fischer received her B.Sc in BiomedicalEngineering after studying at the Technion-IsraelInstitute of Technology and Massachusetts Instituteof Technology (MIT) in 2011 and received her M.Sc inMechanical Engineering in 2013. Currently she is aPhD student at the Biorobotics and Biomechanics Lab(BRML) at the Department of Mechanical Engineeringat the Technion. Her research focuses on gaitbiomechanics with an emphasis on the kinetics,kinematics and electromyography of the lower limbunder the effect of body weight unloading.

Eytan M. Debbi, PhD, graduated from Yale Universitywith a B.Sc in Biology and Biotechnology. He is cur-rently pursuing his Medical Degree at Tel AvivUniversity Sackler School of Medicine and finishedhis PhD at the Biorobotics and Biomechanics Lab(BRML) at Mechanical Engineering faculty at theTechnion-I.I.T. His research interests include biome-chanics, biorobotics, biomedical engineering,biotechnology and musculoskeletal disorders.

Alon Wolf, PhD, earned all his academic degreesfrom the Faculty of Mechanical Engineering atTechnion-I.I.T. In 2002, he joined the RoboticsInstitute of Carnegie Mellon University and theInstitute for Computer Assisted Orthopaedic Surgeryas a member of the research faculty. He was also anadjunct Assistant Professor in the School of Medicineof the University of Pittsburgh. In 2006, he joined asthe Faculty of Mechanical Engineering at Technion,where he founded the Biorobotics and BiomechanicsLab (BRML). The scope of work done in the BRMLprovides the framework for fundamental theories inkinematics, biomechanics and mechanism design,with applications in medical robotics, rehabilitation

robotics, and biorobotics, such as snake robots.