Asymmetries in the Technique and Ground Reaction Forces of Elite Alpine Skiers Influence Their Slalom Performance
Round 1
Reviewer 1 Report
The article is focused on the influence of asymmetry of biomechanical variables on several aspects of alpine skiing slalom. Kinematics and GRF were collected on skiers performing a series of turns starting either on the right or on the left of the track. The symmetry coefficient and Jaccard index were assessed for a combination of independent variables (IV, biomechanical parameters) related to dependant variables (DV, performance variables). Multiple linear regressions were used to evaluate the relationship between IV and DV.
What is the hypothesis, does asymmetry is detrimental or beneficial to the performance? What is the interest to study the asymmetry of performance?
Since whole body kinematics was recorded, did you consider other joint angles than that corresponding to flexion. At the hip, in particular, abduction-adduction could be of interest? Same remark for the trunk and upper-body.
Line 54: “Although the body movements of athletes, and especially those of elite alpine skiers when turning, are often asymmetric” Please define what you mean by asymmetry, it could be asymmetry between the inside and outside leg for example.
Line 95 : “In addition, the skier’s boots were equipped with pressure insoles (Loadsol, Novel GmbH, Munich, Germany) that assessed the total ground reaction force, the individual forces acting on the entire inside and outside foot, and the distribution of force between the fore- and rear-foot.” By total GRF do you mean the 3D vector? Please clarify.
Line 130: “As an indicator of overall (as opposed to average) asymmetries throughout the entire turn, the Jaccard index (JI) [27] was also calculated”. Could you provide some guidelines on what the values mean? For example, does a JI = 50% means little of great asymmetry?
Line 140: “Outliers were excluded from further analysis” On which basis data were considered to be outliers?
Line 150: “… indices for the independent variables (skiing technique and ground reaction forces)”. It does not seem to me that GRF is an independent variable since it is depending on each subject.
Line 155: “as well as a tendency towards a statistically significant difference in flexion of the outside knee (p = 0.11).” The difference is significant or not according to the chosen significant level for p.
Lien176: “The mean angle of the outside shank during left and right turns differed visually during the entire steering phase” What do you mean by visually?
Line 192: “Our multi-variable regression models” Could you please be more precise, what kind of multivariable regression?
Line 194:” (t-statistic, p<0.05) ” Would it be rather p>0.05 ?
Line 198: “SI values for instantaneous and sectional performance” Please define what is instantaneous and sectional performance.
Line 207-208: Table 3: How do you explain the difference in regression models for SI and JI for the same parameter, the turning radius?
Line 225: “ii) asymmetry in turning radius demonstrated the largest predictor coefficients for asymmetry in the angle of the shank and GRF on the entire foot of the outside leg.” If considering SI. If JI is considered the hip flexion seems also to be important.
Line 277: “(e.g., the attack anagle defines as the angle between the longitudinal axis of the ski and the ski’s centre point’s velocity vector projected onto a plane parallel to the surface of the snow) [14])” “angle defined” instead of “anagle defines”
Author Response
The article is focused on the influence of asymmetry of biomechanical variables on several aspects of alpine skiing slalom. Kinematics and GRF were collected on skiers performing a series of turns starting either on the right or on the left of the track. The symmetry coefficient and Jaccard index were assessed for a combination of independent variables (IV, biomechanical parameters) related to dependant variables (DV, performance variables). Multiple linear regressions were used to evaluate the relationship between IV and DV.
What is the hypothesis, does asymmetry is detrimental or beneficial to the performance? What is the interest to study the asymmetry of performance?
Reply to the reviewer: We now include the following formulation: “Our hypothesis was that asymmetries in the performance of elite slalom skiers are influenced by asymmetries in their technique and ground reaction forces.” In practical terms, we believe that changing less effective turns (e.g., left turns) to more closely resemble the more effective turns (e.g., in this example the right turns) should help optimize a skier’s overall performance. However, this remains to be tested employing a different study design.
Since whole body kinematics was recorded, did you consider other joint angles than that corresponding to flexion. At the hip, in particular, abduction-adduction could be of interest? Same remark for the trunk and upper body.
Reply to the reviewer: We did consider other joint angles as well, including all three Euler joint angles at the knee and hip, as well as the inclination of the trunk. Previous findings (Zorko et al. 2013) have demonstrated that the “visual abduction” in terms of 3D joint angles involves true “Euler angle” abduction and a combination of flexion and external rotation (or, more precisely, lower internal rotation). In our opinion, this visual abduction is the observation of most relevance to understanding the biomechanics of alpine skiing. The same is true for the hip angles.
Moreover, angles at the knee joint are particularly complicated by the associated degrees of freedom (Lu et al 2008, Wilson et al 200), so that these angles must be monitored with great care and appropriate statistical analysis and interpretation performed. In addition, it is uncertain whether the differences (asymmetries) in the angles of abduction and rotation are sufficiently large to allow accurate measurement, since with all kinematic systems, including IMUs, the errors involved in measurement of these angles are more pronounced than when determining flexion (the problem of the order in which the Euler angles should be calculated).
Finally, the inclination of CoM in combination with knee and hip joint angles are reflected directly in the ground reaction forces and angle of the shank, which were analysed here. Similarly, most of the inclination of the trunk of alpine skiers is usually reflected in the distribution of pressure between the forefoot and rearfoot, which was also analysed.
These are the primary reasons why we did not analyse the parameters mentioned in this comment by the reviewer.
Line 54: “Although the body movements of athletes, and especially those of elite alpine skiers when turning, are often asymmetric” Please define what you mean by asymmetry, it could be asymmetry between the inside and outside leg for example.
Reply to the reviewer: This sentence has been revised to indicate more clearly that we are focusing on differences between left and right turns.
Line 95 : “In addition, the skier’s boots were equipped with pressure insoles (Loadsol, Novel GmbH, Munich, Germany) that assessed the total ground reaction force, the individual forces acting on the entire inside and outside foot, and the distribution of force between the fore- and rear-foot.” By total GRF do you mean the 3D vector? Please clarify.
Reply to the reviewer: We only measured the component perpendicular to the sole of the ski-boot, as now clarified.
Line 130: “As an indicator of overall (as opposed to average) asymmetries throughout the entire turn, the Jaccard index (JI) [27] was also calculated”. Could you provide some guidelines on what the values mean? For example, does a JI = 50% means little of great asymmetry?
Reply to the reviewer: We now provide some description of what JI means in practice, explaining the two most extreme cases, i.e., when JI = 1 or 0.
Line 140: “Outliers were excluded from further analysis” On which basis data were considered to be outliers?
Reply to the reviewer: Outliers were determined by application of Tukey’s fences, a standard non-parametric procedure for the detection of outliers. With this procedure, a “fence” boundary at a distance of 1.5 IQR (Inter Quartile Range) from the 1st and 3rd quartiles is created and any data outside these fences are considered outliers. This detection is now explained in greater detail.
Line 150: “… indices for the independent variables (skiing technique and ground reaction forces)”. It does not seem to me that GRF is an independent variable since it is depending on each subject.
Reply to the reviewer: In this context, the independent variables in the multivariable linear regression models are those that are considered independent in connection with statistical analysis. At the same time, all descriptive statistical parameters related to the ground reaction forces are expressed as % BW (body weight) and are thereby also independent of the skier’s weight. This is also the case for the indices of symmetry, if this was of concern.
Line 155: “as well as a tendency towards a statistically significant difference in flexion of the outside knee (p = 0.11).” The difference is significant or not according to the chosen significant level for p.
Reply to the reviewer: This part of the sentence has been deleted.
Lien176: “The mean angle of the outside shank during left and right turns differed visually during the entire steering phase” What do you mean by visually?
Reply to the reviewer: This sentence has been revised to describe more precisely what we mean by “visually”.
Line 192: “Our multi-variable regression models” Could you please be more precise, what kind of multivariable regression?
Reply to the reviewer: For added clarity, we have revised this text to read “…linear regression models, each involving no more than 2 predictor (independent) variables,…”
Line 194:” (t-statistic, p<0.05) ” Would it be rather p>0.05 ?
Reply to the reviewer: In this case, p<0.05 is correct, since this p-value represents the t-statistic for the hypothesis that the coefficients of the corresponding model did not differ significantly from zero. In practice, this test was applied to eliminate such models. This sentence has also been revised for greater clarity.
Line 198: “SI values for instantaneous and sectional performance” Please define what is instantaneous and sectional performance.
Reply to the reviewer: We now define instantaneous and sectional performance as suggested, with additional explanation also in the Methods.
Line 207-208: Table 3: How do you explain the difference in regression models for SI and JI for the same parameter, the turning radius?
Reply to the reviewer: These are both associated with the GRF, but the SI for turning radius is associated with the SI for GRF and the JI for turning radius with JI for GRF. Moreover, the JI for shank angle is involved only in the model used to predict the JI for turning radius. Apparently, according to our multivariable regression model, the behaviour (JI) of the shank angle during turning also predicted the behaviour (JI) of the turning radius. In contrast, this was not the case for parameters based on mean values (e.g., SI values for turns). In this context we are unable at present to explain the negative association between the JI for hip flexion and SI for turning radius. Consideration of this aspect of our study in the Discussion has now been expanded upon.
Line 225: “ii) asymmetry in turning radius demonstrated the largest predictor coefficients for asymmetry in the angle of the shank and GRF on the entire foot of the outside leg.” If considering SI. If JI is considered the hip flexion seems also to be important.
Reply to the reviewer: We only list the variables associated with the largest predictor coefficients, therefore leaving out the substantially lower and negative (coefficient) association between hip flexion and turning radius (model #4, Table 3). An additional reason for not including this among our major findings is that at present, we are unable to provide a reasonable explanation for this negative association, which indicates that the less pronounced asymmetry in flexion of the outside hip, the greater the asymmetry in turning radius. We now consider this more extensively in the Discussion.
Line 277: “(e.g., the attack angle defines as the angle between the longitudinal axis of the ski and the ski’s centre point’s velocity vector projected onto a plane parallel to the surface of the snow) [14])” “angle defined” instead of “angle defines”
Reply to the reviewer: Revised as indicated.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments
This study analysed the asymmetries for solemn skiers, and the results are interesting. There are some questions or details to be clarified, especially on methods.
- When 20 HZ and 100 HZ signals were transformed into 240 HZ, were the noise and errors brought in?
- Please briefly introduce the calculation of CoM, and possible errors which should not affect the analysis of asymmetries and mechanical energy.
- What was the definition on energy? Why did the values were negative in Table 2?
- Please briefly introduce the definition for JI.
- How used Box-Cox transformation and which variables were dealt with at this way?
- How did authors define independent or dependent variables in multi-variate regression?
- It is not very clear why the instantaneous and sectional performance was largest predictor. What was predicted by these variables?
Author Response
This study analysed the asymmetries for solemn skiers, and the results are interesting. There are some questions or details to be clarified, especially on methods.
When 20 HZ and 100 HZ signals were transformed into 240 HZ, were the noise and errors brought in?
Reply to the reviewer: Since the data were not filtered prior to synchronization, the answer to this question from the reviewer is yes. All data were first interpolated, synchronized and then smoothed as described in the Methods (the first two sentences in the section »2.3 Computation«). Filtering prior to upsampling and synchronization would negatively affect synchronization (i.e., the detection of signal peaks). In our experience, there is no meaningful difference in the results when the data are smoothed prior to upsampling/synchronization or subjected to appropriate filtering afterwards. These are the reasons we decided to handle our data as described. However, data from GNSS and inertial motion capture were aggregated after filtering. The corresponding sentence in methods was revised to pronounce the aggregation after filtering.
Please briefly introduce the calculation of CoM, and possible errors which should not affect the analysis of asymmetries and mechanical energy.
Reply to the reviewer: We utilized Demster's regression equations (1955) with inclusion of the mass of both skiing and measuring equipment. This approach to calculating CoM can result in errors if the form of the subject’s body differs considerably from that of Demster’s subjects. However, summation of the centre-of-masses of several body segments decreases the impact of error in a single measurement, resulting in a much lower position error. We now describe our calculation of the CoM in greater detail.
What was the definition on energy? Why did the values were negative in Table 2?
Reply to the reviewer: The parameters related to energy dissipation were defined in accordance with the articles cited, in which they were first introduced. In the case of instantaneous performance (differential specific mechanical energy – mechanical energy per unit change in altitude, normalized to the skier’s mass) and sectional performance (mechanical energy for each specific section/turn normalized to the entrance speed), the values of these parameters are negative when energy dissipation (final energy Ef) is lower than the initial (Ei) (-> Ef-Ei<0), which is usually the case in alpine skiing. For greater clarity, additional text concerning this matter has been added to both the Methods and Results.
Please briefly introduce the definition for JI.
Reply to the reviewer: The JI is introduced at the end of the section »2.3 Computation«. To help the reader understand the practical meaning of this index, additional text has been added here.
How used Box-Cox transformation and which variables were dealt with at this way?
Reply to the reviewer: The Box-Cox transformation, programmed by our Matlab routines when needed, turned out not to be executed on the presented data. Therefore, we have deleted the unnecessary text about this transformation.
How did authors define independent or dependent variables in multi-variate regression?
Reply to the reviewer: To fulfil our objectives, the dependent (predicted) variables chosen were related to performance, while the independent (predictor) variables were related to skiing technique and load. Section »2.4 Statistical analysis« has been revised to make this clearer.
It is not very clear why the instantaneous and sectional performance was largest predictor. What was predicted by these variables?
Reply to the reviewer: We assume that this comment refers to the following sentence in the Abstract (or similar statements in the major findings and Discussion): »Asymmetries in instantaneous and sectional performance were found to have the largest predictor coefficients associated with asymmetries in shank angle and hip flexion of the outside leg.« We are referring here to our findings that the largest predictor coefficients were associated with asymmetries in the performance parameters: in the Models #10-13 (Table 3), these predictor coefficients in connection with the multivariable regression models were 3.97, 5.77, 7.16 etc. This is not the same as stating that »the instantaneous and sectional performance was the largest predictor«. Otherwise, your question/concern would be entirely valid: these asymmetries in performance parameters (SI and JI for instantaneous and sectional performance) were dependent (predicted) variables and not independent (predictor) variables.
Round 2
Reviewer 2 Report
ok now
