Skip to main content
SearchLogin or Signup
Accepted
Published onJan 23, 2019DOI

Calculation of Muscle Activity during Race Walking

by Anne Schmitz and Jaclyn Norberg
Published onJan 23, 2019
Calculation of Muscle Activity during Race Walking
·

ABSTRACT

Race walking has been shown to be a lower impact alternative to running. In fact, lower injury rates are reported in race walkers than runners. This may be due to lower joint forces. However, this theory has not been explicitly tested since joint contact forces are difficult to measure. Therefore, these joint forces are calculated using either a serial or cosimulation approach. In both strategies, a detailed joint model and multibody dynamics model are used in tandem to calculate muscle forces. These muscle forces then contribute to the joint contact forces. Computed muscle control (CMC) is an algorithm that utilizes a proportional-derivative controller, static optimization, and forward dynamics to calculate muscle activations, which are used to subsequently calculate muscle forces. The goal of this study was to investigate the viability of using CMC to calculate muscle activations during race walking, with the intent of using these activations in future studies to calculate joint forces. Instrumented motion capture data (kinematics, kinetics, and electromyography (EMG)) of a representative subject was used from a previous study. In this study, subjects walked, race walked, and ran at a self-selected pace across a walkway with embedded forceplates while marker trajectories and EMG were collected. Muscle actuated forward dynamics simulations (i.e. CMC) were created for each movement. The muscle activations resulting from these CMC simulations were compared to the experimentally measured EMG by performing a cross-correlation. The CMC results were fairly accurate across all muscles (gluteus medius, rectus femoris, vastus lateralis, adductor longus, semitendinosus, tibialis anterior, gastrocnemius, peroneus longus) with correlation coefficients greater than 0.5. There was no apparent relationship between movement type and coefficient. Future work is needed to determine if correlation coefficients of 0.5 are accurate enough for studies looking to accurately quantify joint contact forces.

INTRODUCTION

Race walking has grown in popularity as a lower impact alternative to running. In fact, race walking has been shown to provide similar health benefits at a lower injury rate [1]. This difference in rate of joint injury may be due to the decreased impact loads between the ground and foot seen during race walking as compared to running [2]. Impact loads during race walking and running can travel up the kinetic chain to the knee [3][4]. Since increased tibiofemoral loading has been shown to be correlated with knee osteoarthritis [5], race walking likely induces lower joint contact loads in the knee and may explain the lower injury rate compared to running. However, this mechanistic explanation of lower injury rates cannot be experimentally tested without instrumented knee implants [6].

Joint forces can be calculated using either a serial or cosimulation approach [7]. In both strategies, a detailed joint model and multibody dynamics model are used in tandem to calculate muscle forces. These muscle forces then contribute to the joint contact forces [8]. Computed muscle control (CMC) is one tool used in these frameworks to calculate muscle excitations and activations, which are used to subsequently calculate muscle forces [9].

Since muscle forces are the main contributor to contact forces [10], accurate contact forces require accurate muscle forces. These muscle forces are impractical to measure experimentally; thus making them difficult to validate when output from simulations. These calculated muscle forces are typically validated by comparing the calculated muscle activations with experimentally measured electromyography (EMG) data. These activations have been validated for a multitude of movements already in the literature, e.g. single-leg hopping [11], running [12][13], and walking [12][14]. However, race walking has not been considered in the simulation literature. In order to test the theory that race walking induces lower joint contact loads in the knee than running, the validity of calculated muscle activations that result in muscle forces must first be established. Therefore, the goal of this study is to investigate the viability of using CMC to predict muscle activations for race walking. We hypothesize calculated muscle activations during walking, race walking, and running will be similarly, temporally comparable to measured EMG. All three motions are investigated in order to present a wider context for the race walking results since CMC has been used for running [12][13] and walking [12][14] previously.

METHODS

This study is a secondary analysis of data collected for a previous study [15]. Following a protocol approved by an institutional review board (IRB), instrumented motion capture data of an experienced, competitive race walker (mass: 53.2 kg, height: 1.63 m, and age: 31 years) was collected for three motions: walking, race walking, and running at self-selected speeds across a 15 m walkway. The subject walked at 1.25 m/s, race walked at 1.87 m/s, and ran at 3.44 m/s, calculated from a marker placed on the sternum. Ground reaction force data for the stance phase of motion was obtained using two embedded forceplates sampling at 1000 Hz (Bertec Corp., Columbus, OH). Whole body marker trajectories for eighty-five retro-reflective markers were collected at 200 Hz with an 11-camera motion capture system (Motion Analysis Corp., Santa Rosa, CA). A standard marker configuration was used, the details of which are presented elsewhere [15][16]. Electromyography (EMG) (Delsys, Natick, MA) was simultaneously collected for eight muscles: gluteus medius (GM), rectus femoris (RF), vastus lateralis (VL), adductor longus (AL), semitendinosus (ST), tibialis anterior (TA), gastrocnemius (GA), and peroneus longus (PL) using standard SENIAM electrode placements [17][18]. The EMG signals were subsequently postprocessed by creating a linear envelope: (1) low pass, 4th order, zero-lag Butterworth filter with a 100 Hz cutoff frequency, (2) full wave rectified, (3) low pass, 4th order, zero-lag Butterworth filter with an 8 Hz cutoff frequency, and then (4) normalized by maximal voluntary isometric contraction [15][19][20].

Using a Visual3D (C-Motion, Germantown, MD) to OpenSim [21] pipeline [22], muscle actuated forward dynamics simulations were created for each movement (Fig. 1).

<p>Figure 1: Flowchart of the Visual3D to OpenSim pipeline used to create muscle-actuated forward dynamic simulations of movement. These simulations calculate the muscle excitations needed to achieve an experimentally measured motion.</p>

Figure 1: Flowchart of the Visual3D to OpenSim pipeline used to create muscle-actuated forward dynamic simulations of movement. These simulations calculate the muscle excitations needed to achieve an experimentally measured motion.

First, Visual3D was used to process the raw motion capture data (in .c3d format) to produce scaling factors and perform inverse kinematics [15][16]. A description of the bone-embedded reference frames and inverse kinematics algorithm is described in more detail elsewhere [15]. Then, these scaling factors, which were calculated as ratios between the bone lengths in the generic model and bone lengths of the actual subject [23], were input into OpenSim version 3.2. Next, these scaling factors were used to scale the OpenSim generic model gait2392 [21]. The generic model in Visual3D and OpenSim used the same anatomical segments, anthropometrics, and biomechanical joint models. However, the gait2392 OpenSim model additionally contained 92 musculotendon actuators. Then, the residual reduction algorithm (RRA) was used to create a dynamically consistent system (e.g. [24]). RRA creates a dynamically consistent system by slightly changing the mass properties of the bones of the scaled model and the inverse kinematics data so Newton’s equation of motion can be satisfied (ΣF = ma). Since F is force applied to the body as measured via the ground reaction forces (i.e. forceplate data), mass, m, is modeled from anthropometric studies, and acceleration, a, of the body segments is also measured from marker trajectories (i.e. derivative of inverse kinematics), Newton’s equation of motion is often inconsistent due to measurement errors. RRA modifies the m and a variables so the equals sign can be realized. This modified model and marker data is used by the computed muscle control (CMC) tool to calculate muscle forces [9].

CMC first uses a proportional-derivative (PD) controller to calculate the accelerations needed to move a model towards an experimental data trajectory (Eq. 1 [9]):

θ∗¨(t+T)−θp¨(t+T)=kv[θp˙(t)−θ˙(t)]+ku[θp(t)− θ(t)]\ddot{\theta_\ast}\left(t+T\right)-\ddot{\theta_p}\left(t+T\right)=k_v\left[\dot{\theta_p}\left(t\right)-\dot{\theta}\left(t\right)\right]+k_u[\theta_p\left(t\right)-\ \theta\left(t\right)]

In this equation, θ∗¨\ddot{\theta_{*}} are the desired joint angular accelerations calculated after a time interval T; θ̇ and θ are the angular velocity and position of the joints reached by the model due to muscle forces, respectively; the subscript p denotes variables that are the kinematic variables of the prescribed motion (i.e. experimentally measured motion); kv and ku are the velocity and position feedback constants of the PD controller. This means that if the model is behind where the experimental kinematics are at time t, the controller will accelerate the model until t + T and reassess if the model is on track yet. If the model is ahead, the controller will decelerate the joints in the same way. Once the desired joint accelerations are computed, the CMC algorithm integrates muscle activation and contraction dynamics forward from t to t + T for a range of muscle excitations to determine the maximum and minimum forces that can be achieved by the muscles at time t + T. These musculotendon actuators are non-linear force generators whereby the force depends on the length of the muscle, muscle excitation, and the shortening/lengthening velocity of the muscle (i.e. the Hill muscle model [25]). Since the number of muscles is greater than the number of joints, the redundant system is treated as an optimization problem. Therefore, static optimization is used to calculate the musculotendon forces that lie within the previously calculated bounds, produce the desired accelerations from the PD controller, and minimize the summed excitation across all muscles (i.e. a surrogate measure for muscle energy expenditure) [9].

CMC simulations were performed for a representative trial of each experimentally measured movement: walking, race walking, and running. The muscle activations resulting from these CMC simulations were compared to the experimentally measured EMG by performing a cross-correlation in Matlab (MathWorks, Natick, MA). The variables of interest extracted were the maximum value of the correlation coefficient and the phase delay [11][26]. These variables were used to quantify the accuracy of CMC to calculate muscle activations.

To better understand the relationship between movement type and CMC accuracy, a cocontraction index was used to quantify the complexity of each movement. The cocontraction index was calculated as the ratio between the rectus femoris and semitendinosus activity (Eq. 2 [27]):

EMGS/EMGL * (EMGS + EMGL)

In this equation, EMGS is the EMG activity for the less active muscle while EMGL is the activity of the more active muscle. Linear regressions were then used to quantify the relationship between CMC accuracy (i.e. maximum correlation coefficient, phase delay) and movement type (maximum cocontraction index). Cocontraction was used as a surrogate of movement complexity and type because CMC may have difficulty in predicting coordination that involves coactivation. This comes from the optimization step of the CMC framework where the sum of muscle activations squared is minimized. Coactivation, which is clinically considered activation of antagonistic joint pairs [28], would cause an increase in the cost function and hence not be chosen by the CMC algorithm as a preferred coordination pattern.

RESULTS

The maximum cross-correlation coefficient between the calculated muscle excitations and experimentally measured EMG were greater than 0.5 for walking, race walking, and running (Fig. 2). The maximum correlation coefficients averaged across all muscles were 0.5 for normal walking, 0.68 for race walking, and 0.7 for running. The phase delay averaged across all muscles was 22% stance for normal walking, 32% for race walking, and 28% for running. The time traces of the calculated muscle excitations, measured EMG, and cross-correlations can be found in the supplemental information.

<p>Figure 2: The maximum cross-correlation coefficient and phase delay between the calculated muscle excitations and measured EMG for all three movements: normal walking (NW), race walking (RW), and running (RU).</p>

Figure 2: The maximum cross-correlation coefficient and phase delay between the calculated muscle excitations and measured EMG for all three movements: normal walking (NW), race walking (RW), and running (RU).

There was no significant relationship between cocontraction and rectus femoris correlation (p = 0.86), rectus femoris phase delay (p = 0.94), or semitendinosus phase delay (p = 0.61) (Fig. 3). Although, as cocontraction increased, the maximum semitendinosus correlation coefficient did significantly increase linearly (p = 0.01).

<p>Figure 3: The relationship between rectus femoris/semitendinosus cocontraction and the maximum cross-correlation coefficient (i.e. accuracy of simulation to calculate EMG).</p>

Figure 3: The relationship between rectus femoris/semitendinosus cocontraction and the maximum cross-correlation coefficient (i.e. accuracy of simulation to calculate EMG).

DISCUSSION

Joint forces are an important variable of interest when studying joint degeneration, specifically osteoarthritis. To calculate these contact forces, muscle activations and consequently muscle forces must be calculated. Therefore, the accuracy of joint forces is dependent on the accuracy of these muscle activations. Muscle activations can be calculated using a computed muscle control (CMC) algorithm consisting of a proportional-derivative controller, static optimization, and forward dynamics. This framework has been verified by comparing calculated muscle activations to experimentally measured EMG. Although, this has been done for a limited number of movements. This study has expanded on the literature by evaluating the performance of the CMC tool in predicting muscle activations for race walking. Ultimately, the CMC tool calculated activations for race walking in reasonable agreement with EMG, similar to the accuracy found in walking and running (Fig. 2 – Fig. 3).

Overall, the accuracy of the simulations to predict EMG did not vary with movement type (i.e. walking, race walking, and running). This is contrary to suggestions in the literature that the CMC algorithm may have difficulty in predicting cocontraction [12][11]. CMC uses static optimization where the summed muscle activations across all muscles is minimized. This cost function cannot mathematically account for cocontraction. However, the coordination of walking, race walking, and running may not be true clinical cocontraction where cocontraction is typically considered activating antagonistic joint pairs [28]. For example, to produce knee flexion, the CMC algorithm would activate the hamstrings (Fig. 4). If hip flexion accompanies this knee flexion, CMC would also activate the rectus femoris (Fig. 4). This coordination of a biarticular muscle would show up clinically as cocontraction.

<p>Figure 4: (left) To produce knee flexion, CMC would activate the ST. (right) If knee flexion and hip flexion are concurrent in the motion, CMC would activate both ST and RF. This coordination may appear as cocontraction.</p>

Figure 4: (left) To produce knee flexion, CMC would activate the ST. (right) If knee flexion and hip flexion are concurrent in the motion, CMC would activate both ST and RF. This coordination may appear as cocontraction.

Hip flexion and knee flexion simultaneously occur during all three type of gait (walking, race walking, and running) during early stance to loading response and pre- to initial swing phases [29] (Fig. 5). At early stance, a knee flexion moment is produced by the hamstring muscles to absorb the contact forces during initial contact of all three gaits, with less during race walking due to the nature of the gait (i.e. knee extended from heel strike until the body is in full vertical upright position [30]). While this is occurring, a hip flexor moment is created by the rectus femoris to bring the leg forward for initiation of stance phase. Similarly, as the foot comes off the ground (i.e. toe off) during early swing, the hip joint moves from hip extension to hip flexion creating a hip flexion moment and the knee joint moves from knee extension to knee flexion creating a knee flexion moment to prepare for the next contact (i.e. heel strike). These events occur simultaneously where the hamstrings and the rectus femoris, all of which are biarticular muscles, are producing muscle activity at the same time to produce cocontraction.

<p>Figure 5: Hip flexion and knee flexion occur simultaneously during early stance and early swing in walking, race walking, and running.</p>

Figure 5: Hip flexion and knee flexion occur simultaneously during early stance and early swing in walking, race walking, and running.

There are some study design choices to consider when interpreting the results of this study. First, only knee cocontraction was considered. Hip and ankle cocontraction was not considered since the muscles at these joints require indwelling EMG electrodes. This was beyond the scope of the initial study. Second, the same optimization strategy was used for all three gaits where the sum of muscle activations squared (i.e. a surrogate measure of muscle energy expenditure) was minimized in calculating muscle activations. It is unclear if the same coordination strategy is truly utilized during all three movements in vivo; although we would expect the same results if various optimization strategies were used for each movement (i.e. CMC versus static optimization).

In summary, we have used the CMC framework to calculate muscle activations for walking, race walking, and running. The results show these activations agree fairly well with measured EMG, regardless of the motion. We theorize CMC can calculate muscle cocontraction patterns for biarticular muscles, although not for uniarticular muscles since CMC uses a static optimization tool. This is an important first step in determining the usefulness of CMC to be used in simulations calculating joint contact forces for osteoarthritis simulations. The next step is to determine if the correlations found in this study are accurate enough to calculate reasonable joint contact forces during race walking.

NOMENCLATURE

CMC

Computed muscle control. An algorithm, which includes a proportional-derivative controller, used to calculate muscle forces.

EMG

Electromyography. Measurement of muscle excitations.

RRA

Residual reduction algorithm. Used to make dynamically consistent simulations.

GM

Gluteus medius muscle.

RF

Rectus femoris muscle.

VL

Vastus lateralis muscle.

AL

Adductor longus muscle.

ST

Semitendinosus muscle.

TA

Tibialis anterior muscle.

GA

Gastrocnemius muscle.

PL

Peroneus longus muscle.

SUPPLEMENTAL INFO

Experimentally measured EMG and calculated muscle activations are a function of time. To quantify the relationship between these time varying signals, cross-correlation was used. To present a complete picture of the time varying data, plots of the calculated muscle activations, measured EMG, and cross-correlation are included for walking (Fig. S1), race walking (Fig. S2), and running (Fig. S3).

<p>Figure S1: The time-varying data for calculated muscle activations and experimentally measured EMG (top) were compared using cross-correlation (bottom) for normal walking.</p>

Figure S1: The time-varying data for calculated muscle activations and experimentally measured EMG (top) were compared using cross-correlation (bottom) for normal walking.

<p>Figure S2: The time-varying data for calculated muscle activations and experimentally measured EMG (top) were compared using cross-correlation (bottom) for race walking.</p>

Figure S2: The time-varying data for calculated muscle activations and experimentally measured EMG (top) were compared using cross-correlation (bottom) for race walking.

<p>Figure S3: The time-varying data for calculated muscle activations and experimentally measured EMG (top) were compared using cross-correlation (bottom) for running.</p>

Figure S3: The time-varying data for calculated muscle activations and experimentally measured EMG (top) were compared using cross-correlation (bottom) for running.

License
Creative Commons Attribution 4.0 International License (CC-BY 4.0)
Comments
74
?
Login to discuss
jm
jane man: pretty dope analysis ezcardinfo
Devin Berg: After a review of the comments of the peer reviewers and the responses and edits of the authors, I have decided to accept this pub for inclusion in The Journal of Open Engineering. I would like to thank the reviewers, Dr. Sritharan and Dr. Modenese, for their thoughtful and thorough reviews. I would also like to thank the authors for their responses and edits. I encourage the authors to monitor the comments on this page as future readers may provide additional useful insights into this work.
AS
Anne Schmitz: “4. I concur with the other Reviewer’s statements on co-contraction. The Jinha et al papers are important as they further dispel the idea that static optimisation cannot predict co-contraction in uniarticular muscles.” Optimization does not mathematically predict co-contraction.
AS
Anne Schmitz: In the above manuscript, I have revised the CMC explanation to include “the CMC algorithm integrates muscle activation and contraction dynamics forward from t to t + T for a range of muscle excitations to determine the maximum and minimum forces that can be achieved by the muscles at time t + T. These musculotendon actuators are non-linear force generators whereby the force depends of the length of the muscle, muscle excitation, and the shortening/lengthening velocity of the muscle (i.e. the Hill muscle model).” The static optimization step will not predict co-contraction for an agonist pair of muscles at the same length-velocity. Therefore, the above step where the contraction dynamics are used to calculate the limits of the muscles is likely where co-contraction comes from.
AS
Anne Schmitz: “3. I suggest that kinematics and kinetics of race walking should be presented in the Supplementary Material as not all readers would be familiar with them.” This is beyond the scope and aims of this secondary study. These results can be found in the primary publication, Norberg’s thesis.
AS
Anne Schmitz: “2. There is still no discussion with regards to the comparison of EMG with the CMC results. I note that the other Reviewer has also referred to this their feedback. I had mentioned in my original review that despite good cross-correlation findings for some muscles, some temporal patterns clearly differ between EMG and CMC. The cross-correlations presented as main findings essentially mask the actual visually identifiable differences in the temporal patterns presented as supplementary material. The authors would need to provide some discussion on this. I noted in my previous feedback that the results in the Supplementary Material would need to be moved into the main Results in order to facilitate this.” The only way to quantify the agreement of the temporal patterns is through cross-correlation. Visual comparison would be subjective at best. Therefore, I have decided to keep the cross-correlation results.
AS
Anne Schmitz: “1. It is still not clear whether the co-contraction index was calculated as a simple ratio (RF/ST or ST/RF) or whether the equation of Palmieri-Smith (2009) was used, which is actually the equation proposed by Rudolph et al. (2000), in which is not a simple ratio.” The co-contraction index was calculated according to Palmieri-Smith (2009) equation. The manuscript has been edited to clarify the equation.
PS
Prasanna Sritharan: Thank you for undertaking this revision. I think the manuscript, mainly the Methods section, is certainly improved. However, there are some issues that remain, detailed below.   1. It is still not clear whether the co-contraction index was calculated as a simple ratio (RF/ST or ST/RF) or whether the equation of Palmieri-Smith (2009) was used, which is actually the equation proposed by Rudolph et al. (2000), in which is not a simple ratio.   2. There is still no discussion with regards to the comparison of EMG with the CMC results. I note that the other Reviewer has also referred to this their feedback. I had mentioned in my original review that despite good cross-correlation findings for some muscles, some temporal patterns clearly differ between EMG and CMC. The cross-correlations presented as main findings essentially mask the actual visually identifiable differences in the temporal patterns presented as supplementary material. The authors would need to provide some discussion on this. I noted in my previous feedback that the results in the Supplementary Material would need to be moved into the main Results in order to facilitate this.   3. I suggest that kinematics and kinetics of race walking should be presented in the Supplementary Material as not all readers would be familiar with them.   4. I concur with the other Reviewer’s statements on co-contraction. The Jinha et al papers are important as they further dispel the idea that static optimisation cannot predict co-contraction in uniarticular muscles.
AS
Anne Schmitz: “There is a ‘GA’ in the plots missing from the nomenclature.” This was meant as the abbreviation for gastrocnemius. The manuscript has been corrected.
AS
Anne Schmitz: “This comment from my previous review needs to be addressed as well: “Also the collected EMG data should be assessed against some previous literature, at least for walking [3] and running [4] to ensure that the gold standard used to assess CMC results is actually a high-quality EMG recording.”. You should discuss explicitly how the EMG signals presented in your study compare to previous studies. In your current plots, EMGs for normal walking do not look like previous literature for VL, AL, ST, TA. These signals look flat in the plots and some of them have activation equal to one (VL and AL), which is unrealistic. Can you justify these results, or some measurements were corrupted?” The collected EMG was validated in Norberg’s thesis where the data was initially collected.
AS
Anne Schmitz: 1.      “CMC uses static optimization where the summed muscle activations across all muscles is minimized. This cost function cannot mathematically account for cocontraction.” this is not true. Static optimization can predict co-contraction in 3D joints even for monoarticular muscles, as this has been demonstrated analytically in several papers from Walter Herzog’s group (see [1,2] as examples). 2.      “We theorize CMC can calculate muscle cocontraction patterns for biarticular muscles, although not for uniarticular muscles since CMC uses a static optimization tool.”. Again, static optimization can predict cocontraction for uniarticular muscles in three-dimensional models, as proved mathematically in Jinha, A., et al., Antagonistic activity of one-joint muscles in three-dimensions using non-linear optimisation. Mathematical Biosciences, 2006. 202(1): p. 57-70. Fundamentally, CMC is not formulated to calculate co-contraction of agonistic muscle pairs. “Antagonistic muscle activity can … increase metabolic cost…” - Knarr et al, (2012) J EMG and Kines, 22(4). When the cost function of muscle activity squared is minimized as used by the static optimizer of CMC, which is practically minimizing muscle energy expenditure, muscle co-contraction (i.e. antagonistic muscle activity) will be minimally used, if even at all.
LM
Luca Modenese: Thank you for your revision, I think that the manuscript, and especially the Methods section, now reads much better than in the previous version. However, now there is more emphasis on contact forces during race running, which makes the manuscript’s aim less direct than before. I would slightly refine the aims in the Introduction once again, emphasising the novelty of estimating muscle force in race walking more than its role as intermediate step towards joint contact force estimation for studying osteoarthritis and injury, application that requires other approaches to model validation. Also, you did not reply directly to some of my comments from the previous review. I will reformulate them in a more explicit manner. The following sentences about the static optimization objective function are incorrect and need revision: 1.      “CMC uses static optimization where the summed muscle activations across all muscles is minimized. This cost function cannot mathematically account for cocontraction.” this is not true. Static optimization can predict co-contraction in 3D joints even for monoarticular muscles, as this has been demonstrated analytically in several papers from Walter Herzog’s group (see [1,2] as examples). 2.      “We theorize CMC can calculate muscle cocontraction patterns for biarticular muscles, although not for uniarticular muscles since CMC uses a static optimization tool.”. Again, static optimization can predict cocontraction for uniarticular muscles in three-dimensional models, as proved mathematically in Jinha, A., et al., Antagonistic activity of one-joint muscles in three-dimensions using non-linear optimisation. Mathematical Biosciences, 2006. 202(1): p. 57-70. This comment from my previous review needs to be addressed as well: “Also the collected EMG data should be assessed against some previous literature, at least for walking [3] and running [4] to ensure that the gold standard used to assess CMC results is actually a high-quality EMG recording.”. You should discuss explicitly how the EMG signals presented in your study compare to previous studies. In your current plots, EMGs for normal walking do not look like previous literature for VL, AL, ST, TA. These signals look flat in the plots and some of them have activation equal to one (VL and AL), which is unrealistic. Can you justify these results, or some measurements were corrupted? The conceptual distinction between “true” clinical co-contraction and model co-contraction/coordination from the Discussion section remains unclear. You defined co-contraction as ratio between RF and ST activities (last paragraph of Methods), so whenever the model activates both these muscles there is muscle co-contraction by definition. The net joint moments causing this model muscle recruitment strategy (or coordination, as named in the manuscript) offer an explanation for the co-contraction, do not make it less “true”, so I cannot understand your argument in the Discussion section. Please add details, at least in the reply, if not in the paper. There is a ‘GA’ in the plots missing from the nomenclature. REFERENCES 1. Jinha, A., et al., Antagonistic activity of one-joint muscles in three-dimensions using non-linear optimisation. Mathematical Biosciences, 2006. 202(1): p. 57-70. 2. Jinha, A., R. Ait-Haddou, and W. Herzog, Predictions of co-contraction depend critically on degrees-of-freedom in the musculoskeletal model. Journal of Biomechanics, 2006. 39(6): p. 1145-1152.
LM
Luca Modenese: Could you please provide answers to the comments in a more structured way as you would normally do in a manuscript revision and give indications about what you have changed in response to what? Thank you.
LM
Luca Modenese: This is a first review where I made some comments following the headers suggested in the journal guidelines. GENERAL COMMENTS This manuscript focuses on a well-defined methodological question, i.e. if the computed muscle control (CMC) algorithm can accurately predict muscle activations when used to simulate different activities. Shapes of activation signals and co-contraction are considered, assessed through cross-correlations with EMG signals and co-contraction indices of semitendinosus (ST) and rectus femoris (RF) respectively. A standard model (gait2392) and workflow (inverse kinematics followed by residual reduction analysis (RRA) and CMC) are employed. The study concludes that CMC can produce fairly accurate results for the considered motions. In my opinion the research question could be of some interest for the modelling community, but the manuscript requires substantial improvements in many aspects. TECHNICAL SOUNDNESS In my opinion not all the parts of the manuscript are technically sound. The description of the CMC algorithm in the Methods section, referring to the induced acceleration analysis, is not correct. This sentence is wrong: “The muscle excitations needed to achieve the overall joint accelerations is determined using the induced acceleration results and by minimizing the summed excitation across all muscles.”. Muscle activation squared is minimized in CMC using static optimization, as correctly stated at the end of the manuscript, and induced acceleration is not used to calculate the joint acceleration, which his just derived from a standard forward dynamic simulation (see https://simtk-confluence.stanford.edu:8443/display/OpenSim/How+CMC+Works). In my opinion, this sentence is also incorrect: “CMC uses static optimization where the summed muscle excitations across all muscles is minimized. This cost function cannot mathematically account for cocontraction.” as static optimization in itself can predict co-contraction in 3D joints even in monoarticular muscles, as this has been demonstrated analytically [1, 2]. Same for this sentence from the conclusion: “We theorize CMC can calculate muscle cocontraction patterns for biarticular muscles, although not for uniarticular muscles since CMC uses a static optimization tool.”. The last part of the sentence is incorrect for the same reasons. Also the collected EMG data should be assessed against some previous literature, at least for walking [3] and running [4] to ensure that the gold standard used to assess CMC results is actually a high-quality EMG recording. Finally, the experimental data were processed with Visual3D and an MSc thesis is referenced to explain the scaling of the model, so making unclear what the effective procedure was for readers not familiar with Visual3D. No scaling factors, nor tracking errors, nor pelvis residuals are reported in the paper. I would include them to help the reader evaluating the quality of the simulations. CLARITY It is unclear what would be the difference between co-contraction observed in EMG (“true” co-contraction) and the muscle strategy computed by CMC recruiting RF and ST, which is referred to “coordination” of biarticular muscles showing up as co-contration. What would should the model provide, in order for its muscle activations to be considered “true co-contractions”? COMPLETENESS In my opinion the method section lacks the details of the experimental collection are not reported, and a PhD thesis is referenced instead. I would add them. OPENNESS AND REPRODUCIBILITY There are no models or simulations made available with the publication. The workflow is standard, so the part relying on OpenSim in principle should be reproducible if the experimental data used for the study were made available in some form. REFERENCES 1. Jinha, A., et al., Antagonistic activity of one-joint muscles in three-dimensions using non-linear optimisation. Mathematical Biosciences, 2006. 202(1): p. 57-70. 2. Jinha, A., R. Ait-Haddou, and W. Herzog, Predictions of co-contraction depend critically on degrees-of-freedom in the musculoskeletal model. Journal of Biomechanics, 2006. 39(6): p. 1145-1152. 3. Perry, J., Gait Analysis: Normal and Pathologic Function. 1992, Thorofare, NJ USA: SLACK Incorporated. 4. Hamner, S.R., A. Seth, and S.L. Delp, Muscle contributions to propulsion and support during running. Journal of Biomechanics, 2010. 43(14): p. 2709-2716.
+ 4 more...
PS
Prasanna Sritharan: Did the use of the stance phase rather than the whole gait cycle affect the results?
AS
Anne Schmitz: Swing data was not collected in the initial study since impact loading is the main focus of this line of inquiry. I have altered the Introduction to better motivate race walking and looking at impact during this motion.
PS
Prasanna Sritharan: I’m confused as to what this set of results is intended to show. There is no elaboration provided in the Discussion.
AS
Anne Schmitz: Thank you for the comment. The Methods have been reworded to better justify this analysis.
PS
Prasanna Sritharan: Given the numerous issues identified, the Abstract will require substantial revision.
AS
Anne Schmitz: Thank you for the suggestion.
PS
Prasanna Sritharan: The manuscript title “Effect of Motion Type on Muscle Activity Calculations” is too broad and vague. Please consider revising this to provide more information to the reader.
AS
Anne Schmitz: Thank you for the suggestion.
PS
Prasanna Sritharan: Are the authors reporting activation or excitation? These are related but different quantities. The caption states “excitations” but the figures report activation. This applies to all the supplementary results.
AS
Anne Schmitz: Thank you for the comment. I have revised the captions to read activation.
PS
Prasanna Sritharan: As previously mentioned, to the naked eye, the patterns of most CMC activations do not correspond with EMG, even taking into account a small time delay. Furthermore the EMG for some muscles is flat, suggesting noise or malfunction. This comment applies to all the supplementary results. Can the authors please clarify the validity of these results?
AS
Anne Schmitz: Thank you for your comment.
PS
Prasanna Sritharan: This is not entirely true since CMC incorporates excitation-activation-contraction dynamics, in theory it can predict greater co-contraction than static optimisation in many cases, especially in ballistic tasks where the muscle force demand can change rapidly, but the muscle cannot respond fast enough due to the transients present in its excitation-contraction dynamics.
AS
Anne Schmitz: Thank you for your comment.
PS
Prasanna Sritharan: As previously mentioned, I strongly disagree with this based on: (1) the magnitude of phase delay; and (2) visually comparing the EMG and CMC muscle activation waveforms in the Supplementary Material.
AS
Anne Schmitz: Thank you for your comment.
PS
Prasanna Sritharan: This part of the sentence is difficult to comprehend. Could the authors please clarify what they mean here, or rephrase this segment?
AS
Anne Schmitz: Thank you for the comment. This has been revised.
PS
Prasanna Sritharan: Just a comment: activation here is used as a surrogate measure for metabolic cost – it is not the same thing.
AS
Anne Schmitz: Thank you for your comment. This has been revised.
PS
Prasanna Sritharan: What is the key message of this narrative paragraph with respect to the results of the present study? Also, the Discussion focusses heavily on co-contraction while the primary aim of the study is to compare “accuracy” of CMC for various movement types. I understand that the type of movement has been quantified by the level of co-contraction, however, as previously noted, I feel this is not a good way to delineate the task type.
AS
Anne Schmitz: Thank you for your comment.
PS
Prasanna Sritharan: What is a true co-contraction? I would argue that any time two muscles are activating, they are co-activating (or co-contracting), as per the example given by the authors in Figure 4. Can the authors please clarify this statement? Why would the example of Figure 4 not be considered a true co-contraction?
AS
Anne Schmitz: Thank you for your comment. An explanation of cocontraction has been added.
PS
Prasanna Sritharan: I think the authors have fundamentally misinterpreted how CMC works. The static optimisation module within CMC is used to generate a steady-state future force set which is then used to estimate excitations that are then input into a forward dynamics model. This forward dynamics model takes into account excitation-activation-contraction dynamics therefore has the potential to predict co-contraction better than, say, static optimisation.
AS
Anne Schmitz: Thank you for your comment.
PS
Prasanna Sritharan: As previously mentioned, I strongly disagree with this statement. Lin et al. do not suggest this anywhere in their work.
AS
Anne Schmitz: Thank you for your comment.
PS
Prasanna Sritharan: Again, I do not think it is appropriate to quantify the movement type or complexity by the level of co-contraction.
AS
Anne Schmitz: Thank you for your comment.
PS
Prasanna Sritharan: Are you reporting excitations or activations? The supplemental figures report activations.
AS
Anne Schmitz: This has been changed to activations.
PS
Prasanna Sritharan: I feel this is a misleading statement based on: (1) the magnitude of phase delay; and (2) visually comparing the EMG and CMC muscle activation waveforms in the Supplementary Material. Please see my overall comments with respect to the Results.
AS
Anne Schmitz: Thank you for your comments.
PS
Prasanna Sritharan: I’m not sure that “varying complexity” is an appropriate description here. I would argue that walking is more “complex” than running as it requires more complicated neuromuscular coordination than running. Also as previously noted, the novelty of this work is unfortunately diminished somewhat by a previous study by Lin et al. (2012) who also compared CMC results for running and walking, and also compared CMC results with other computational techniques for those tasks.
AS
Anne Schmitz: The Introduction has been reworded to better emphasize the race walking aspect of the study.
AS
Anne Schmitz: The Discussion has been revised to reflect this comment.
PS
Prasanna Sritharan: The authors should make it clear to the reader that CMC is just one of many methods that can be used to calculate muscle activations.
PS
Prasanna Sritharan: Unfortunately the Discussion is quite poorly written and points to difficulty in interpreting the results beyond “EMG and CMC results agree” (which I disagree with). In particular it is difficult to appreciate (1) what the key findings of this study are; and (2) how these results will benefit modelling studies in future. I would suggest the authors first address the considerable issues with the Methods and Results, and then perhaps revisit Discussion. A revised set of results may provide more insightful interpretation. Furthemore, two key parts of the discussion are not provided: (1) the authors have not compared the results of their work – correlations, and temporal patterns of muscle forces and EMG – with previously reported results; (2) a discussion of the major limitations of their work.
AS
Anne Schmitz: Thank you for your comment. The Discussion explicitly states how this study expands the current literature. Also, a paragraph on study design considerations (i.e. limitations) is also included.
PS
Prasanna Sritharan: Provide p-values to 3 sig. figs even if not significant.
AS
Anne Schmitz: P-values were reported to 2 decimal places to be consistent with traditional statistical reporting methods. Field, A. (2013). Discovering statistics using IBM SPSS statistics. sage.
PS
Prasanna Sritharan: Just a comment: It is important to note that the equation of Rudolph et al. provides no information as to which muscles are more highly activated.
AS
Anne Schmitz: Thank you for the comment.
PS
Prasanna Sritharan: I’m not sure “movement difficulty” the best term to describe the increased physical effort required from walking to race walking to running. They are different tasks that require increasingly more physical effort, but are they increasingly more “difficult” tasks? I would argue, on purely anecdotal evidence of me trying to race walk, that race walking is a significantly more “difficult” task than either running or walking, which are more “natural”.
AS
Anne Schmitz: Thank you for the comment. The wording has been removed.
PS
Prasanna Sritharan: Given that three different tasks were performed, please provide time histories of the joint angles and possibly also joint torques for walking, race walking and running as either main results or supplementary material.
AS
Anne Schmitz: These detailed results can be found in the original study, Norberg PhD thesis.
PS
Prasanna Sritharan: Does a cross-correlation analysis provide p-values? If so, please include them in the results.
AS
Anne Schmitz: There are no p-values to report.
AS
Anne Schmitz: The figures are activations. The captions have been updated.
PS
Prasanna Sritharan: In the supplementary figures, are the authors reporting activation or excitation? These are related but different quantities. The caption states “excitations” but the figures report activation. I feel the supplementary material needs to be moved in the main results. This is because an important part of the Discussion would be to compare the authors results with those of other studies, to effectively validate their work against any previous results.
PS
Prasanna Sritharan: This is a major concern with regards to the Results. Are these phase delays reasonable? Approximately how many seconds (or milliseconds) does, for example, 22% stance represent? Phase delays significantly beyond a reasonable time lag for excitation-contraction dynamics would suggest that, in fact, the results have limited meaning within the context of muscle activations. In other words, if the temporal distance between correlating phase segments is large, the correlation is unlikely to be meaningful in the present context. Can the authors provide some comment on the magnitudes of the phase delays reported? It seems from the supplementary results, that the time histories of muscle activations for a large number of the presented muscles do not, to the naked eye, reasonably match the temporal patterns of the EMG, even taking into consideration small time delays for excitation-activation-contraction transients. This applies to all three tasks. Where good correlations seem to visually occur, the phase delay seems to be large. This is corroborated by the quantitative results presented. Therefore any correlation may simply be coincidence. Furthermore, for walking, the EMG profiles of some muscles are flat suggest excessive noise or malfunctioning electrodes. Can the authors please explain these large qualitative differences, and also the flat EMG profiles of some muscles, such as VL, ST, TA and AL in walking?
AS
Anne Schmitz: Thank you for your insight.
PS
Prasanna Sritharan: If more than one trial were used, then please report the results as means and standard deviations.
AS
Anne Schmitz: Thank you for the comment. Only 1 representative trial was used.
PS
Prasanna Sritharan: Does the use of the stance phase rather than the whole gait cycle affect the results?
AS
Anne Schmitz: Swing data was not collected in the initial study since impact loading is the main focus of this line of inquiry. I have altered the Introduction to better motivate race walking and looking at impact during this motion.
PS
Prasanna Sritharan: Please include subject demographic data (age, sex weight, height, etc) as well as trial data (walking/running speed, etc) here or in the Methods section.
AS
Anne Schmitz: As suggested, this has been added to the Methods.
PS
Prasanna Sritharan: I have significant concerns with respect to the Results of this study. The objective was to compare CMC results for walking, race walking and running, however, the results focus exclusively on Please refer to my overall comments regarding this.
AS
Anne Schmitz: Thank you for your comments. The Introduction and Methods have been reworded to better motivate the study.
PS
Prasanna Sritharan: The authors have quantified the movement type based on co-contraction index. What is the justification for doing this? What evidence is there that complex tasks demand greater co-activity of antagonist muscles? While the results seem to show that the co-contraction index increases from walking to race walking to running, how the authors know to expect this a priori? I.e. why did the authors choose to quantify the movement type using the co-contraction index – how could they be sure that the transition from walking to running would cause an increase in co-contraction? E.g. studies have found no relationship between “effort” and co-contraction for some tasks, e.g. drop jump (Kellis et al. 2003)
AS
Anne Schmitz: Thank you for the comment. This was done to better understand the differences seen in cross-correlation results. The manuscript has been edited.
PS
Prasanna Sritharan: Please indicate the level of statistical significance, e.g. α = 0.05)
AS
Anne Schmitz: Thank you for the comment.
PS
Prasanna Sritharan: The authors of the present study define the co-contraction index as a ratio of RF and ST activity. However, they cite Palmieri-Smith et al., who in fact use the definition of co-contraction index by Rudolph et al., which is not a simple ratio. Can the authors please clarify how they calculated co-contraction index, and include the equation used, and appropiate reference, e.g. Rudolph et al.
AS
Anne Schmitz: Thank you for the comment. The co-contraction index was calculated as a ratio, which is stated in the manuscript.
PS
Prasanna Sritharan: The authors only compute co-contraction across the knee for RF and ST muscles, which are both bi-articular. (1) Why did the authors only choose these biarticular muscles? (2) What about the other knee-spanning muscles? Would it be more appropriate to include the vasti and also the other hamstring muscles? Also, please see next comment.
AS
Anne Schmitz: Thank you for the comment. This knee pair is the most commonly used when clinically evaluating cocontraction to stabilize the knee.
PS
Prasanna Sritharan: It seems from the results that cross-correlation was only performed for stance. This may have strongly affected the results. I feel that cross-correlation would have been better applied to the full gait cycle. This way the analysis is being performed on cyclic data, with the start and end of the cycle corresponding to the same point in the cycle. Thus during the cross-correlation, as the convolving waveform is time shifted and wraps around the other, one full cycle is covered.
AS
Anne Schmitz: Thank you for your insight. Only stance data was collected for the initial study. Also, stance is usually most clinically relevant.
AS
Anne Schmitz: More details of the EMG postprocessing have been added.
PS
Prasanna Sritharan: How was the EMG processed? E.g. rectified, filtered, linear enveloped, etc. Please provide details of cut-off frequencies etc.
PS
Prasanna Sritharan: Why is race walking an important task? No motivation was provided for this in the Introduction.
AS
Anne Schmitz: Thank you for the comment. The Introduction has been revised.
PS
Prasanna Sritharan: How many trials were performed per tasks? How many trials were selected to generate the results per task, and what criteria were used to select/discard trials?
AS
Anne Schmitz: Only 1 trial was used for each motion. I have revised the wording to make this clearer.
PS
Prasanna Sritharan: Please provide a reference for this, if appropriate
AS
Anne Schmitz: Thank you for the recommendation. I have added the Thelen article here as a reference.
PS
Prasanna Sritharan: The description of CMC is somewhat oversimplified, in that it misses a fundamental component: static optimisation. The authors have essentially described CMC as simply a feedback controller, whereas this is not the whole story. CMC uses a combined static and dynamic framework. At each time step, static optimisation is used to estimate the required muscle forces necessary to achieve the required accelerations, which are then used to find the appropriate excitations. These excitations are then input into the model and integrated forward (Thelen & Anderson 2006). I would strongly suggest adding more detail here.
AS
Anne Schmitz: Thank you for the suggestion. I have revisited Thelen’s article and revised the wording to be more precise.
PS
Prasanna Sritharan: What is the purpose of this reference? To show how RRA is used? If so, this is unnecessary. Just reference the generic OpenSim paper by Delp et al.
AS
Anne Schmitz: This reference presents an example of using RRA while the OpenSim paper by Delp does not.
PS
Prasanna Sritharan: These sentences are poorly constructed. I assume you mean the skeletal linkage/topology of the OpenSim model was the same as that of the Visual3D model?
AS
Anne Schmitz: Thank you for the suggestion. More details have been added.
PS
Prasanna Sritharan: More detail is required regarding the model. While this generic model is familiar to OpenSim users, it will not be true for others. Please describe: (1) the number of segments, joints and muscles; (2) the number of degrees of freedom of each joint and/or the type of kinematic constraint used for each joint, e.g. ball, pin, etc.; (3) the type of muscle model used, e.g. Hill Type.
AS
Anne Schmitz: Thank you for the suggestion. More details have been added.
PS
Prasanna Sritharan: The reference to the Norberg study is by itself insufficient. Please describe this in more detail.
AS
Anne Schmitz: The Norberg study is a multi-year, doctoral dissertation that details the collection and processing of the motion capture data. A detailed review of this methodology would be overly long for a simulation study and detract from the overall goal, which is the computational aspect.
PS
Prasanna Sritharan: What version of OpenSim was used?
AS
Anne Schmitz: Thank you for the comment. The version has been added.
PS
Prasanna Sritharan: Please briefly state what Visual3D and OpenSim are, for the benefit of those unfamiliar with either or both. E.g. OpenSim is an open-source musculoskeletal modelling package, etc.
AS
Anne Schmitz: Thank you for the suggestion
PS
Prasanna Sritharan: What systems were used for motion capture, ground force measurement and EMG? Please include manufacturer details. What marker protocol was used for the gait analysis?
AS
Anne Schmitz: More details have been added.
AS
Anne Schmitz: Thank you for your insight. The Introduction has been re-worded to better emphasize the novelty of the race walking motion.
PS
Prasanna Sritharan: Lin et al. already examined walking and running using CMC, which tempers the novelty of this work. Race walking presents a new task not previously examined. However, all three are sagittal-plane movements. Perhaps in future the authors could consider frontal-plane or pivoting movements. This would be interesting from the point of view of knee injuries, OA and ACL-reconstruction.
PS
Prasanna Sritharan: No detail is provided of the participant recruitment process or experimental protocols. And no demographic detail is provided of the representative subject. Also, the reference here is a PhD thesis, which is probably not appropriate. If this experimental data has, since submission of this present work, been published as a research article elsewhere please update this reference. Otherwise, I feel you will need to fully detail the participant recruitment process, experimental protocols, lab set up and equipment manufacturers, and also provide any additional information, including human ethics approvals if required by the journal. Please include the demographic details of the representative participant, e.g. age, sex, weight, height, etc. Please show the speeds at which the participant walked, race walked and a ran. It would also be prudent to show that human ethics approval was obtained for this work - however this would be at the discretion of the journal. Is the subject and experienced race walker and/or runner? How would the biomechanics of an inexperienced race walker differ from experienced ones? Why was race walking chosen? Do race walkers have high rates of knee OA?
AS
Anne Schmitz: Thank you for the suggestion. We have added more details of the data collection process to stay consistent with prior analyses of the biomechanical data. Below are examples of how we have published secondary analyses of data and summarized the data collection process: Schmitz, A., Norberg, J. (2018) Frontal Plane Biomechanics during Walking. Current Trends in Biomedical Engineering and Biosciences. 12(1). doi: 10.19080/CTBEB.2018.12.555826 Schmitz, A., Norberg, J. (2016) The Sensitivity of Joint Torques during Running to Forceplate Data Error. ASME Journal of Dynamic Systems, Measurement and Control Special Issue on Biomedical Sensing, Dynamics, and Control for Diagnostics, Treatment, and Rehabilitation. 138(11): 111001-1. In addition, all of these comments are addressed in more detail in the PhD thesis. The thesis is searchable through Google Scholar and available through the University of Kentucky, hence this is an appropriate citation. Since this work is a secondary analysis of previously collected data, a full, detailed description of the data collection process would be redundant and inappropriate as it would suggest the data was collected for this specific work.
PS
Prasanna Sritharan: Insufficient detail is provided in the section, and hence it is not possible to effectively reproduce the results of this work.
AS
Anne Schmitz: Thank you for the comment. More detail has been added.
PS
Prasanna Sritharan: I’m nit-picking again, but this statement, while nicely rounding off the Introduction, is probably unnecessary. More words/space could be dedicated to further explaining/evaluating the CMC method and its limitations/challenges.
AS
Anne Schmitz: Thank you for the suggestion. This statement has been removed.
PS
Prasanna Sritharan: This objectives statement is too broad. The movement tasks should be listed, and the metric used to determine accuracy should also be stated, e.g. by comparing with patterns of recorded EMG. A more detailed discussion of the limitations/difficulties associated with the CMC method may facilitate the development of a nice hypothesis, e.g. that the running results may be less accurate than walking because…
AS
Anne Schmitz: Thank you for the comment. The goal of the study has been re-worded.
PS
Prasanna Sritharan: The phrase “…these studies were conducted independently…” is not correct. The Lin et al. paper compared running and walking within the same study. This study concluded that CMC results were consistent with EMG timing and on/off sequence, although they did not compare the temporal shape of the EMG curves. Unfortunately, for my mind, this diminishes the novelty of the present work, which also examines CMC-based muscle activation solutions for running and walking. Please refer to my overall comments for a further elaboration of this.
AS
Anne Schmitz: Thank you for your comment. The Introduction has been significantly re-worded to emphasize the novelty of using CMC to calculate muscle forces during race walking.
PS
Prasanna Sritharan: I strongly disagree with this statement based on the authors’ supplied reference. Lin et al. in no way claims that “accuracy” decreases with increased “motion complexity”. Lin et al. suggested that muscle activations in ballistic tasks or those with time-dependent criteria may be less accurately predicted by methods that do not take into account activation dynamics. Ballistic tasks are not necessarily “more complex” than less ballistic tasks. E.g. the coordination of muscles in walking is inherently more complex than a maximal vertical jump. In any case, Lin et al. were comparing the pros and cons of various computational approaches for predicting muscle forces, therefore I would assert that their statements re “accuracy” are not appropriate for this present work. One principle limitation of CMC is that it cannot be used to examine time-dependent criteria because it utilises static optimisation within the framework, however it also does utilise forward dynamics which means it can potentially predict co-contraction better than static optimisation alone. In fact, in some ballistic tasks, CMC would certainly predict co-contraction better due to the inclusion of excitation-activation-contraction dynamics. If torque demand changes quickly during a task, the muscle forces may not be able to keep up, which may result in elevated co-contractions
AS
Anne Schmitz: Thank you for your insight. This sentence has been removed from the Introduction.
PS
Prasanna Sritharan: This sentence does not read well. I would consider revising, e.g. “A commonly-used method for validation of the accuracy of calculated muscles forces to compare against experimentally-measured EMG data”. Care must be taken with regards to this statement. Are the authors saying that EMG is used to validate both temporal patterns and magnitude? I would suggest extending this sentence to include a note that using EMG to establish accuracy has limitations. E.g. using EMG to compare magnitudes of activations may not be appropriate in painful/pathological cohorts, or where a good reference EMG (e.g. max volunatary isometric contraction) cannot be obtained. Furthermore, to better link this sentence to the remainder of the paragraph, perhaps indicate that not only are the individual temporal activation patterns typically compared against the EMG, but also, assuming the use of EMG in the examined cohort is appropriate, it can also be used to determine levels of co-activation between muscle groups (e.g. Zeni et al. 2010).
AS
Anne Schmitz: Thank you for the comment. I have reworded this sentence for clarity.
PS
Prasanna Sritharan: I would probably support this statement with one or more appropriate references. Perhaps Pandy & Andriacchi (2010) is a nice general reference which supports this, however any studies that report muscular contributions to joint forces could potentially be appropriate here.
AS
Anne Schmitz: Thank you for the suggestion.
+ 1 more...
PS
Prasanna Sritharan: I’m nit-picking here, but this is not entirely true. I would say “…CMC is one technique used in these frameworks…”. A range of computational strategies have been applied in both sequential and co-simulation approaches, including various forms of dynamic optimisation (Shelburne et al. 2005, Lloyd & Besier 2003, etc.), static optimisation (Brandon et al. 2014, Sritharan et al. 2012, etc.), and more recently collocation techniques (Koelewijn & van den Bogert 2016).
AS
Anne Schmitz: Thank you for this insight.
+ 1 more...
PS
Prasanna Sritharan: It is interesting to note that CMC may be overtaken by collocation-based techniques which are becoming popular in computational biomechanics and is well-suited to tracking problems.
AS
Anne Schmitz: Thank you for that insight.
AS
Anne Schmitz: Thank you for the suggestion. The Introduction has been edited to better motivate the race walking motion and tie it to osteoarthritis.
PS
Prasanna Sritharan: The link between osteoarthritis and the three analysed tasks – walking, race walking and running – is never established.
PS
Prasanna Sritharan: I would suggest this first paragraph, while providing the biological/clinical motivation for the study, is overly long. The first sentence of the Discussion actually summarises this paragraph quite neatly, and I would suggest using that in place of this paragraph, and devoting more space and words of the Introduction to better establish the modelling-based motivations for this study.
AS
Anne Schmitz: Thank you for the suggestion. I have replaced the biological motivation with a rationale for studying race walking.
PS
Prasanna Sritharan: Unfortunately, the Introduction it not well written and provides limited or no motivation for the present study, particularly as Lin et al. (2012) has already covered some of the main objectives.
AS
Anne Schmitz: Thank you for the comment. The next study in this line of inquiry aims to compare contact forces between race walking and running to determine if race walking is lower impact and hence a “safer” exercise, at least from a cartilage load standpoint. The initial study here in TJOE is a step towards this goal: can CMC be used to calculate reasonable muscle forces for race walking? If these are not reasonable to begin with, the resulting contact forces will also be unreasonable. The Introduction has been altered to better motivate this study with the next work in mind. Next work: Schmitz, A. and Norberg, J. (2019) Effect of Motion Type on Contact Forces. (abstract, paper, and oral presentation for the American Society of Mechanical Engineers International Mechanical Engineering Congress and Exposition). Salt Lake City, UT: ASME IMECE.
The Journal of Open Engineering
Published with