Vibration Attenuation Via Mean of Lower Limb Muscles Occurs During Whole Body Vibrations And Differs Across Frequencies And Postures

17 Lower limb muscles actively contribute to maintain body posture but also act to attenuate soft 18 tissues oscillations that occur during everyday life. This elicited activity can be exploited as a 19 mean of neuromuscular training or rehabilitation. In this study, Whole Body Vibrations (WBV) at 20 different frequencies were delivered to healthy subjects while holding static postures to test the 21 transient muscles mechanical responses. Twenty-five participants underwent WBV at 15, 20, 25 22 and 30 Hz while holding either a static ‘ hack squat ’ or ‘ fore feet ’ posture. Soft tissue 23 accelerations and surface electromyography (sEMG) were recorded from Gastrocnemius 24 Lateralis (GL), Soleus (SOL) and Tibialis Anterior (TA) muscles. Estimated displacement at muscle 25 bellies revealed a resonant pattern, different across frequencies and postures (p<.001). 26 Specifically, a peak in the displacement was measured after the onset of the followed by a drop and a further plateau (only after few seconds after the peak) suggesting a 1 delayed neuromuscular activation. Although oscillation dampening was correlated to an 2 increased muscular activity, only specific WBV settings were promoting a significant muscle 3 contraction. For example, SOL and GL induced activation was maximal for subject in forefeet 4 and while exposed to higher frequencies (p<.05). The non-immediate response of leg muscles 5 to a vibratory stimulation confirms the tonic nature of the vibration induced muscle contraction 6 (the tonic vibration reflex) and its strong influence on postural tonic muscles (GL and SOL). This 7 may have significant impact on training or rehabilitation protocols aiming towards postural and 8 balance improvement or recovery.

followed by a drop and a further plateau (only after few seconds after the peak) suggesting a 1 delayed neuromuscular activation. Although oscillation dampening was correlated to an 2 increased muscular activity, only specific WBV settings were promoting a significant muscle 3 contraction. For example, SOL and GL induced activation was maximal for subject in forefeet 4 and while exposed to higher frequencies (p<.05). The non-immediate response of leg muscles 5 to a vibratory stimulation confirms the tonic nature of the vibration induced muscle contraction 6 Introduction 12 Whole Body Vibration (WBV) refers to the use of mechanical stimulation, in the form of vibratory 13 oscillations extended to the whole body, to elicit neuromuscular responses in multiple muscle 14 groups 1 . Vibrations are generally delivered through lower limbs via the use of platforms on 15 which subjects stand. When WBV was included in training and rehabilitation programmes, 16 physical exercises were performed on such platforms 2 . This approach has become increasingly 17 popular as it evokes a large muscle response and, more importantly, it elicits muscles activity 18 through physiological pathways via the Tonic Vibration Reflex (TVR) mechanism 3 , improving the 19 overall motor performance while enhancing strength and flexibility [4][5][6][7][8][9] . 20 The TVR has been proven to explain an increased and synchronised motor-unit (MU) firing rates 21 recorded during locally-applied (i.e., focal) vibrations 10,11 . Indeed, when vibrations are applied 22 directly to tendons or muscle bellies, muscle fibres length changes activating a reflex response 23 from muscle spindles. This translates in an increased MU firing rates phased-locked specifically 24 to the vibratory cycle, i.e. the TVR 10,12,13 . 25 Although in WBV vibrations are not applied locally, they are transferred to the target muscles 26 via the kinematic chain determined by the body posture [14][15][16] . This provides similar muscular 27 outcomes with respect to focal stimulations as well as additional systemic postural responses, 28 allowing better flexibility and applicability to large exercise programmes 4 . Specifically, when the 29 whole body is exposed to mechanical shocks (such as vibrations), absorption strategies act to 30 dampen oscillations and dissipate energy through modulation of both muscle activity and joint 31 kinematics, over which the body has prompt control 17,18 . Moreover, in WBV, somatosensory 1 feedback pathways are enhanced by reflexes arising from mechanoreceptors in the lower limbs, 2 with significant implications for motor coordination and postural control during quiet stance 19 . 3 Although promising results of WBV training are reported in the literature 20-28 , a few discording 4 results still jeopardize the systematic use of such approach in training and rehabilitation 5 practices 29-31 . Conflicting results might be related to the high amount of variability in WBV 6 settings (e.g., stimulation frequency, posture, stimulation amplitude, stimulation duration etc.) 7 used throughout different studies, while still lacking of standardised training protocols. Among 8 the most investigated variables, both stimulation frequency and subject posture have relevant 9 impact in eliciting an efficient muscle tuning response to WBV 32-34 . Previous findings suggest 10 that muscles contract to reduce the soft-tissue resonance, especially when the stimulation 11 frequency, ω a , is close to their natural one 35-37 . This process, known as muscle tuning, is 12 perpetrated by muscles to minimize the soft-tissue vibrations 38,39 and has been recently 13 proposed as one of the possible body reactions to WBV 40,41 . Therefore, a careful selection of 14 stimulation frequency, ω a , to match the resonant one, ω 0 , seems the key element to maximise 15 muscle responses to WBVs 42 . Generally, the natural frequency of a system depends on its mass, 16 , and stiffness, according to the formula ω 0 =√k/m 4 . While the mass of a muscle can be 17 considered as a constant, its stiffness can be modulated by muscle activation in a given body 18 posture. Changes in subjects' posture do therefore change muscles' stiffness, therefore leading 19 to a change in the muscles' natural frequency. During dynamic exercises on a vibrating platform 20 the body kinematic chain involved in the transmission of the mechanical stimulus changes 21 continuously, making it difficult to define the stimulus delivered at the target muscle group. 22 During static WBV exercises instead, the energy dissipated through joint kinematics is constant 23 and muscle contraction is the major mechanism tuned to dampen vibration. Abercromby et al. 24 did in fact confirm that static exercises during WBV enhance muscle response than performing 25 dynamic exercises, during which muscles contract in an eccentric and concentric fashion 40 . 26 However, no actual physiological justification has been provided on the reason why static WBV 27 exercises might be more efficacious than dynamic ones. 28 We hypothesised that muscles would require an intrinsic time interval to react to the vibratory 29 stimulation (TVR response) and, based on a given stimulus (e.g. frequency) and body posture, to 30 tune muscle stiffness accordingly. In addition, we hypothesised that the extent of vibration 31 dampening is related to the increase of muscle activity and viceversa. To test our hypothesis, we 32 recorded and analysed muscle displacement -derived from accelerometers placed on muscle 1 bellies-and muscle activation in response to WBVs delivered via a side alternating platform at 2 different frequencies while the subjects held different static postures. 3

4
Participants and experimental design 5 Seventeen females and eight males (age: 24.8 ± 3.4 years; height: 172.0 ± 8.6 cm; mass: 64.6 ± 6 10.5 kg) volunteered in the study after providing written informed consent. History of 7 neuromuscular or balance disorders as well as recent injuries were among the exclusion criteria. 8 To evaluate muscle activation and displacement during WBV, surface electromyography (sEMG) 9 signals and accelerations were collected from three lower limb muscles during two static 10 exercises performed in static conditions (without WBVhereafter called baseline activity) and 11 when different vibration frequencies were delivered. The protocol of the study received platform (Galileo ® Med, Novotec GmbH, Pforzheim, Germany), as it was shown to evoke bigger 30 neuromuscular activations than synchronous vibrating ones 32 : a peak-to-peak amplitude of 4 1 mm was used. For each subject, ten trials were collected to evaluate the effect of five stimulation 2 frequencies that covered the frequency range offered by the platform -0, 15, 20, 25, 30 Hz-and 3 two subject postures: hack squat (HS) and fore-foot (FF). To ensure heels off the ground during 4 the FF trials, subjects were asked to keep their heels in contact with a parallelepiped-shaped 5 foam (30 x 4 x 3 cm) glued on the platform while keeping their lower limb straight. During HS 6 trials instead, subjects were asked to keep their knees flexed at about 110° and a goniometer 7 was used to check the angle at the beginning of each HS trial. Trials were administered in a 8 random order with a one-minute break between consecutive trials. 9 Hereafter, trials with vibratory stimulation are referred to as the "WBV trials" Natick, MA). Accelerations were band-pass filtered between 10 and 100 Hz to remove gravity 25 components and accommodation movements, usually confined between 0 and 5 Hz 46,47 , and to 26 retain only vibration-induced muscle displacements, located mostly at the stimulation frequency 27 and its superior harmonics 48 . Filtered epochs were then double integrated to estimate local 28 displacement along the different axes ( , , ) and the total displacement recorded 29 at each muscle level was estimated as: 6 (t) = √ ( ) 2 + ( ) 2 + ( ) 2 1 where = 1,2, … , , with being the total number of samples. 2 To track the low-frequency mechanical muscle response to WBVs, a moving average of 3 ( ) was calculated using a 250ms sliding window (Fig. 1). To compare 4 and superimpose muscle displacement among different subjects, vectors 5 were time-locked to the point where a 0.1 change in the slope was detected, which will be 6 hereafter referred to as the vibration onset, and used for statistical analyses. To describe muscle 7 response to vibrations, two time points were defined as follows: Once a was identified for all muscles and conditions, was defined as: 18 where ( ) is the longest peak duration observed across muscles and 20 conditions. Since the longest peak duration was of 3.12 s, which was recorded for the 21 GL in 30 , was located 4.7 s after the vibration onset (green asterisk in Fig 1). 22 To quantify the extent of the displacement attenuation at each muscle site, 23 was calculated for each subject as the difference between the maximum displacement recorded 24 at that site and the steady-state one: 25

1
To isolate the muscle activity preceding the stimulation ( ) from the one actually induced 2 by the vibrations ( ), each WBV trial was split into two epochs: 10 and 30 seconds, 3 respectively. The central portions of these signals (6 and 20 seconds, respectively) were 4 extracted and retained for analyses. Similarly, the central 20 seconds of the baseline trials ( 0 5 and 0 ) were extracted and retained for analyses. 6 All epochs were band-pass filtered between 5 and 450 Hz with a 5 th order Butterworth filter and 7 a mean running root mean square ( ) value was obtained from both the baseline 8 ( ) and the epochs ( For each muscle, a cluster-based permutation test was used to compare the mechanical 5 response of muscles over time 50,51 for: 6  between the two postures at the four frequencies (four tests); 7  between frequency pairs in HS and FF (twelve tests). 8 Time series comparisons were performed over the portion of the signals between the vibration 9 onset and to include both the peak and stabilization phase and because no effect was 10 expected before the WBVs. 5000 permutations were used to build the random distribution 11 against which the test statistic of the actual signal were compared. An alpha level of 0.05 was 12 used to identify the significant clusters for each comparison 52 . To overcome the multiple To relate the mechanical response with the physiological one, a Pearson correlation 1 coefficient was calculated between and the outlier-free 2 population, after the subjects that were identified as outliers for ANOVA analyses were removed 3 from the respective population. For each muscle, the data recorded in the eight trials 4 ( 15   20  25  30  15  20  25 30 ) were pooled together. 5

6
All subjects were able to undergo WBV stimulations while holding the prescribed postures. The 7 average ankle angles measured in Fore Feet were -9.4° ± 6.4° where a negative measure 8 indicates a plantar flexion. When participants underwent the WBVs in Hack Squat, the average 9 knee angle was 70.8° ± 4.4°. 10 Muscle dynamics analysis 11 Our results confirmed that the muscles dynamics differed significantly depending on the posture 12 and frequency: overall, a larger displacement was observed in HS trials and at lower frequencies. showed a similar trend: a peak with a successive drop and a further stabilisation after some 17 seconds (Fig. 2). 18 More in detail, although peak heights seemed stable across the different explored frequencies, 19 the drop changed significantly among muscles and postures (Fig. 2). GL displacement after the 20 peak, was significantly smaller at higher frequencies -at 20, 25 and 30 Hz rather than at 15 Hz 21 (p=.0002) and at 30 Hz rather than at 20 (p=.0014) (Fig. 2, a.1) in HS. In FF, a similar trend was 22 recorded: a smaller displacement was found at 30 Hz with respect to 25 Hz (p=.0004) (Fig. 2,  23 a.2). 24 The average displacement recorded at the SOL site was smaller at 30 Hz than at 15 (p=.0006), 25 The mechanical response of TA also confirmed the trend observed for the other two muscles: 1 its displacement was always smaller at higher frequencies in HS ( 15 < 20 , p=.0004; 2 25 < 30 , p=.0006; for the other comparisons, p=.0002) (Fig. 3, c.1) and in FF (p=.0002) 3 (Fig. 3, c.2). 4

Muscle activity analysis
5 Normality was confirmed for the dependent variables in most of the conditions, for all three 6 muscles.
was not always normally distributed for TA, but the latter 7 distributions were similarly skewed to those that met normality. Four subjects were removed 8 from the dataset of GL and SOL, and three from that of TA, since 9 represented outlier values. Distribution of values for the different muscles, 10 posture and frequencies is depicted in Fig. 3. 11 A significant WBV-induced muscle activation ( ) was observed in all 12 conditions for the GL (see first row of Table 1) and in most of the conditions for the SOL, apart 13 from 15 (see second row of Table 1). Instead, the TA showed a significant response to WBVs 14 only for 15 and 30 Hz and 25 (see third row of Table 1)

Relation between muscle dynamics and muscle activity
1 A positive correlation was found between the increase of SOL muscle activity and the amount 2 of displacement attenuation (rho=0.2886, p<.001, see Fig. 4, B). No significant correlations were 3 found between the augmented activation of GL and TA and the extent of displacement reduction 4 measured at the respective site (Fig. 4, A and C). 5

Discussions 6
To the authors knowledge, this is the first study analysing the dynamics of the mechanical 7 response of muscle tissues to WBVs and to correlate it with EMG activity to highlight an 8 immediate or delayed response to steady stimulations. Some studies have related the platform 9 acceleration to muscle activity 55,56 , others investigated the relationship between muscle 10 activation and body joint acceleration 34,57,58 and others studied the transmissibility of vibrations 11 to the shank and thigh segments in relation to muscle activity 59,60 . In the latter, a single body 12 posture was used, and the analyses focussed only on the central part of the WBV trials, leaving 13 out the analyses of the initial response to the stimulation. No other study was found to analyse 14 the progressive dynamics of the displacement at the muscle site and EMG activation while 15 undergoing WBVs with different static postures. 16 Our analysis of the dynamics of the displacement and EMG recorded at each muscle site 17 confirmed our hypotheses: muscle reaction to WBVs depends on stimulus characteristics and 18 subject's posture and develops in time to reduce muscle oscillations. Indeed, a common 19 mechanical pattern, never highlighted before, can be observed from our results (Fig. 3). In 20 response to vibratory stimulations, the extent of oscillations of muscles shows a rising phase, a 21 peak oscillation and a subsequent drop, all of which completed within 4 to 5 seconds after the 22 vibration onset, followed by a sustained stable oscillation (plateau). Neither the stimulation 23 amplitude nor the posture of participants varied during individual tests, hence a neuromuscular 24 response is accounted for the observed dynamics. This interpretation aligns to the muscle tuning 25 theory, whereby soft-tissue oscillations arising in response of impact forces applied to the feet 26 are dampened by an increase in muscle activation 37,38,61 . During WBVs, in fact, vibrations are 27 transferred from the feet to the muscles via the body kinematic chain and produce soft-tissue 28 compartment oscillations at the stimulation frequency, which in our case was in the range of the 29 natural frequencies of calf muscles 35 . In light of the reported theory, it is therefore reasonable avoid damage, creating the characteristic raising and falling curves observed in our recordings. 1 The differences observable in these curves confirm that mechanical response changes across 2 muscles, frequencies and posture, suggesting that it is not of artefactual nature but that it 3 actually reflects an underlying muscular activation. Moreover, they also suggest that not all 4 combinations of frequencies and postures can elicit a resonant response in some muscles. accelerations at the knee joint were found to peak at 15 Hz and to dramatically decrease with 10 increasing frequencies, suggesting the occurrence of muscle tuning 34 . Similarly, vibration 11 transmission to the triceps surae and thigh muscle compartments were found to consistently 12 decrease with increasing frequencies 59 , suggesting that a damping effect was more present at 13 frequencies that are closer to the muscles' resonant ones (the higher ones). These results are in 14 line with what we observe in the plateau phase of the mechanical response, nevertheless, these 15 conclusions were drawn on partial information analyses (the central interval of the WBV trials), 16 and do not include the analysis of the initial dynamics. 17 Our study advance the understanding of muscles reaction to WBVs according to stimulation 18 characteristics and, specifically, highlight the tonic nature of muscle reaction to vibration. 19 Indeed, only after an intrinsic interval, which in this study is around 5 seconds, this reaction can 20 completely settle. This may also explain why static exercises (postures) are found to be more 21 effective than the dynamic ones while on vibrating platforms 40 : during the first, muscles can 22 tune to WBVs as opposed to a continuously changing kinematic chain, with changing in muscle 23 contraction and sensitivity to vibrations 11 . 24 In addition, the analysis of the physiological response of muscles to WBVs highlighted specific 25 combinations of posture/frequency able to produce maximal results. As expected from 26 acceleration analyses, also muscles activation varied: GL sEMG activity was significantly 27 enhanced in all WBV combinations, while only specific combinations were effective for SOL and 28 TA activation. This highlight the importance of the selection of appropriate WBV parameters 29 combinations to activate target muscles. In addition, undergoing WBV stimulation while in Fore 30 Feet was found to lead to a higher increase of GL and SOL sEMG activity rather than in Hack responsive 11 , and in our case GL and SOL , both plantar-flexors, are more engaged in FF than HS 1 62,63 . 2 WBVs delivered at 30 and 25 Hz triggered a greater activation in both muscles, as similar findings 3 reported 64 , supporting previous proposal of GL natural frequency residing between 25 and 30 4 Hz 42 . These conclusions are further confirmed by the observation of the permutation test 5 results. Most differences were appreciable for the plateau phase, where the displacement of GL 6 and SOL soft-tissue compartments was significantly reduced at 30 Hz than at other WBV 7 frequencies, further supporting the claim that this frequency is the one triggering the largest 8 tuning effect. Moreover, the positive correlation found between the SOL sEMG increase and the 9 displacement attenuation further suggest that the reduction of displacement in the plateau 10 phase is indeed the manifestation of a neuromuscular response, potentially activated to reduce 11 resonance. The absence of correlation in GL and TA might be explained by the sub-population 12 separation visible in the first and the absence of variance in one of the two population visible in 13 the latter. 14 The absence of any posture or frequency effect on the TA activation during WBVs might be 15 explained by the following: (i) the stimulation frequencies used in this study that were limited 16 to 30 Hz and not enough close to TA's natural frequency, which ranges up to 50 Hz 36 ; (ii) the 17 selected postures that did not lead to an appropriate level of TA engagement, limiting its 18 response to WBVs 11 ; (iii) the phasic nature of the TA, which makes it physiologically different 19 from the other muscles included in this study 65 . 20 Combining the above, it can be inferred that 30 Hz-Fore Feet might be the best combination of 21 stimulation frequency and subject posture when aiming to effectively enhance both GL and SOL 22 muscular responses. For the explored combinations, instead, the TA muscle showed that WBVs 23 elicit muscular activity but did not allow to identify any combination producing a significantly 24 higher response. Therefore, a wider range of frequencies and postures or a completely different 25 approach should be explored. 26 For further studies on the topic, synchronisation of EMG recordings, soft tissue and platform 27 accelerations should be carefully considered and justified. Vibration propagation does in fact 28 depends not only on the level of stiffness of muscles, but also on the gender and 29 anthropometrics of the subjects 66 . With the procedure adopted in this study, it was possible to oscillate (rather than on the platform onset). This allowed a more appropriate synchronisation 1 of muscle activity and mechanical response between subjects. 2 In addition, although the WBV frequencies investigated in our study encompass the range 3 commonly used in WBV training 59 , future studies should expand the investigation to higher 4 frequencies. 5

Conclusions 6
Our results highlighted a muscle driven mechanical response in muscles undergoing vibratory 7 stimulation: a clear trend with a resonant peak followed, after few seconds from the start of the 8 stimulus, by a more stable plateau that reflects a "delayed" neuromuscular activation to modify 9 the properties of the biomechanical system (e.g. muscle stiffness). The non-immediate response 10 of leg muscles to a vibratory stimulation confirms the tonic nature of vibration-induced muscle 11 contraction and its strong influence on postural tonic muscles (GL and SOL).