Simar Gill1*,Emily Sara Matijevich1, Nesa Keshavarz Moghadam1, Sandro Nigg1, Benno Nigg1
1 Faculty of Kinesiology, University of Calgary, AB, Canada
Abstract
Forward acceleration is an important factor to control in biomechanics studies investigating walking/running, as human kinetics and kinematics vary with changes in acceleration. During overground running studies, acceleration is typically estimated using the net anterior-posterior ground reaction force impulse, as measured with force plates. However, as studies move outside the lab, there is limited validation of alternative methods of quantifying acceleration when force plates are unavailable. The purpose of this study was to assess the validity of alternative methods for estimating acceleration in overground running. We evaluated the use of timing lights and motion capture, as indicators of acceleration, to measure change in velocity. We hypothesized that the change in velocity calculated from timing lights and motion capture markers near the center of mass would have a strong positive correlation with the relative anterior-posterior ground reaction force impulse measured with force plates. Ten participants ran in an indoor lab while measurements were collected using timing lights, motion capture and a force plate. The correlation between the relative anterior-posterior impulse and velocity changes measured by timing lights was weak (r = -0.01, r = 0.27 and r = 0.15, respectively). In contrast, the correlation between the relative anterior-posterior impulse and velocity change determined by motion capture was strong (r = 0.81). In studies where force plates are not available, measuring changes in velocity with motion capture is a promising method for calculating and controlling acceleration. However, measuring changes in velocity with timing lights does not show as much promise due to weak correlation values and should therefore be avoided.
Introduction
Biomechanical studies are typically conducted in a laboratory setting, using high-cost, non-portable measurement equipment to evaluate human movement. However, the influx of low-cost wearable and portable measurement equipment in recent years has enabled the conducting of biomechanical studies to be outside the lab, allowing for human move-ment to be evaluated in realistic and unconstrained settings (Ruder, et al. 2019). The transition of biomechanical studies outside the lab requires certain methodologies to be reevaluated in response to these new settings. Similarly, when certain equipment and methodologies are unavailable in lab based studies, replacement methods must be developed and validated.
One important variable to control in biomechanical studies that investigate walking and running is the participant’s forward acceleration. An individual “accelerates” when their forward velocity increases over time, “decelerates” when their forward velocity decreases overtime and is at “steady-state” when there is no change in forward velocity over time. Researchers are often interested in controlling the forward acceleration because kinematics and kinetics change when the forward acceleration changes. For example, reductions in ankle work output were seen in a deceleration state compared to an acceleration or steady-state (Williams et al. 2017). Additionally, another example, there was greater knee extension and plantar flexion velocity in an acceleration state compared to steady-state running (Van Caeken-berghe et al. 2013). To control for the state of motion, vali-dated methods to quantify individuals’ forward acceleration are necessary. Furthermore, when small deviations from constant speed occurred, such as increases in speed, there were changes in step length and time (Rodman and Martin 2020). This variability in steps indicates that it is not possible to maintain the same energetic output during movement, as changes in step length change knee extensor moments, ankle plantar flexor moments and hip flexor moments (Gill et al. 2023).
In a laboratory setting, walking/running biomechanics studies are often conducted on a treadmill or overground. On a treadmill, experimental conditions such as running velocity and acceleration can easily be manipulated by the experimenter, offering greater control over forward velocity and acceleration. However, treadmills, especially expensive, force-instrumented treadmills, may not be readily available in all labs. Furthermore, the speed and acceleration of the treadmill belt is subject to fluctuations (Willwacher et al. 2021) which could result in deviations from the desired velocity/acceleration state.
For some laboratory studies, evaluating overground running may be preferred or even necessary. Overground running may more closely mimic “real-life” scenarios, as individuals spend more time walking/running overground compared to on a treadmill (Yang and King. 2016). Furthermore, some motion analysis labs may have an in-ground force plate available, but not a force-instrumented treadmill. Researchers often debate whether treadmill or overground walking/running protocols are realistic or valuable. A limitation to conducting studies using overground running is that velocity and acceleration cannot easily be manipulated by the experimenter. As such, additional methods are necessary to control forward velocity and acceleration. However, the various methods available for quantifying forward acceleration have not been systematically validated.
To the authors’ knowledge, the anterior-posterior (AP) ground reaction force is the only signal that has been leveraged for measuring and controlling for forward acceleration during overground locomotion. Over a given gait cycle, according to conservation of momentum, the net AP (braking/propulsion) impulse is equivalent to the net acceleration in the forward direction over a stance phase. During steady-state running, individuals run at a constant velocity, and the braking and propulsion force impulses are equal. Specific approaches for selecting steady-state running trials include: (a) identifying gait cycles that fall within a net AP boundary (e.g., -0.045-0.045 m/s2 during the stance phase (Van Caekenberghe et al. 2013); (b) identifying gait cycles with the smallest net AP impulse, through calculation (Bruening et al. 2018) or visual inspection (Williams et al. 2017). As such, we consider the relative AP ground reaction force impulse to be the “gold standard” method for quantifying forward acceleration. The relative AP ground reaction force impulse is the ratio of the positive impulse over the absolute value sum of the positive and negative impulse and is used to estimate the acceleration state (Equation (2)). When force plates are not available, some researchers have controlled for steady-state running by verbally cueing participants to maintain a steady forward acceleration (Havas et al,. 2000; Riley et al. 2008). Surprisingly, many overground running biomechanics studies either have made no effort to control forward acceleration or did not report on the method for controlling or measuring forward acceleration (Breine et al., 2013; White et al. 1998). This raises a broader debate on the validity of kinetics and kinematics reported from overground locomotion studies when it is unknown if forward acceleration was controlled for.
Timing lights and motion capture (optical, IMU-based, markerless) are alternative measurements for quantifying kinematics. Both of these methods include portable options that may be available to researchers to use outside the lab. Both timing lights (Brown et al. 2014) and motion capture (Bruening et al. 2018) have been used to measure walking/running velocity, but researchers do not typically report using this equipment to measure forward acceleration. The validity of these methods for quantifying forward acceleration has not been established. Therefore, the primary purpose of this study is to compare acceleration determined from timing lights and from motion capture with acceleration determined from the relative AP impulse measured with force plates. The results of this study raise a debate on whether timing lights and/or motion capture should be used in place of the gold-standard method when multiple measurement options are available. Since both timing lights (Brown et al. 2014) and motion capture (Bruening et al. 2018) have previously been used to measure walking/running velocity, the agreement between these two methods should be explored. If an agreement cannot be found, further research should be conducted on the accuracy of these two methods. Therefore, a secondary purpose of this study is to compare the relationship between the forward velocity measured with timing lights and the forward velocity measured with motion capture.
The hypothesized results are:
H1: The change in velocity measured with timing lights has a strong positive correlation with relative AP ground reaction force impulse.
H2: The change in velocity measured with motion capture has a strong positive correlation with relative AP ground reaction force impulse.
H3: There is a strong positive correlation between velocity measured with timing lights and velocity measured with motion capture.
Methods
Subjects
Ten participants (eight male and two female) were recruited for the study (mean ± SD, age = 23.8 ± 5.88 years, height = 174.5 ± 6.69cm, weight = 72.2 ± 9.04kg). All participants were injury free (no self-reported injuries or complaints of pain with respect to running). Each participant performed overground running in a laboratory setting. Timing lights, motion capture and force plate data were collected. Footwear and foot strike pattern were not specified. All participants provided informed written consent, and the protocol was approved by the University of Calgary’s Conjoint Health Research Ethics Board (REB21-1069).
Experimental Protocol
Participants performed ten overground running trials for each of the following conditions: (a) Steady-state running, (b) accelerating running and (c) decelerating running. Participants received verbal instructions before each condition informing the participant of the condition and whether to maintain a steady running speed (steady-state), to speed up (accelerating) or to slow down (decelerating) when crossing the capture volume. For all trials, participants were allowed to self-select their running speed as the study’s purpose was to evaluate methods for estimating running acceleration, which could be achieved at any given running speed. Individuals were allowed practice runs to familiarize with the condition and were given rest between trials. Participants were allowed to take as many practice runs as necessary to feel comfortable with the experimental task. No specific instructions were given during the practice runs to accelerate, decelerate or maintain the same speed. Ground reaction forces were collected from a force plate (2400Hz, Kistler Instrument AG, Winterthur, CH). Optical motion capture kinematics were collected using Vicon (200Hz, Nexus 2.11, VICON Motion Systems Ltd, UK). Timing light data were collected with four pairs of timing lights (Brower Timing Systems, Draper, UT, USA). Marker trajectories and ground reaction forces were filtered with a third-order, dual-pass, Butterworth Filter at 8Hz and 20Hz respectively. The ground reaction force cut-off frequencies were based on prior biomechanics running research (Day et al. 2021).
Estimating forward acceleration with timing lights
Four pairs of timing lights were set up across 2.7m distance at a height of 109 cm. Each timing light was equally distanced at 0.9m apart, with the two interior timing lights set up on either edge of the force plate (Figure 1). Using the known distance between pairs of timing lights, and by measuring the time the runner took to pass through each pair of timing lights, average velocities 1, 2 and 3 were calculated (Figure 1). Forward acceleration was estimated as the change between two values of average velocity. The timing light set up was positioned before, at and after the single gait cycle, as measured by both the force plate and motion capture. Therefore, with the use of multiple timing lights and multiple average velocities, a potential estimation of acceleration can be made.
Figure 1: Timing lights set up. Dots indicate placement of timing lights, one set of timing lights are indicated by two dots in vertical alignment. Vavg indicates average velocity measured between two sets of timing lights.
Estimating forward acceleration with motion capture
Motion capture data were collected with four markers placed on the pelvis. These were placed on the anterior superior iliac spine and the posterior superior iliac spine bilaterally via palpation). The average position of these four markers was used as an approximation of the runner’s center of mass kinematics (Vanrenterghem et al. 2010). Seven motion capture cameras were placed circularly around the data collection distance, and these locations were consistent across trials and participants. The derivative of the anterior-posterior component of the center of mass position was taken to obtain the approximate center of mass velocity. To approximate the center of mass forward acceleration over a single stance phase, the difference between forward velocity at toe off and foot contact was calculated (Equation (1)). Foot contact and toe off was determined using a 20N vertical ground reaction force threshold. As part of the secondary analysis, the average center of mass velocity over the stance phase was also calculated, indicating the average stance phase velocity.
$$\text{Center of Mass Acceleration} = \text{velocity}_{\text{toe off}} - \text{velocity}_{\text{foot contact}} \tag{1} $$
Force Plate
One force plate was set up in the center of the data collection volume and used to collect the ground reaction forces of the participants. The ground reaction forces were used to determine the relative AP impulse, which is the ratio of the positive impulse over the absolute value sum of the positive and negative impulse (Figure 2; Equation (2)). When the change in velocity is 0 (steady-state), the positive and negative impulses are equal and opposite with a relative AP impulse of 0.5. Relative AP impulse values greater than 0.5 indicate an acceleration state, while relative AP impulse values less than 0.5 indicate a deceleration state. The calculation of the relative AP impulse is preferred over other methods as it is not speed dependent, allowing for interpretation across a variety of speed ranges. Relative AP impulse was chosen because our participants ran at a range of self-selected running speeds. Consequently, we took a modified approach to the method of quantifying the net AP impulse for estimating acceleration state. We sought a method that would not be confounded by absolute running speed and an acceleration metric that could easily be compared across participants running at different speeds. Relative AP impulse was our gold standard measure of forward acceleration.
Figure 2: Ground reaction force in body weights versus time in seconds. The positive impulse is the area under the curve when the ground reaction force is positive. The negative impulse is the area under the curve when the ground reaction force is negative. B.W. = body weights.
$$\text{Relative AP Impulse} = \frac{\text{Positive Impulse}}{|\text{Negative Impulse}| + \text{Positive Impulse}} \tag{2}$$
Calculations
For each change in velocity calculation, the correlation with the relative AP impulse (gold standard) was determined. To address the secondary purpose of this study, we found the correlation between average velocity 2 (timing lights) and the average velocity determined by the average center of mass (motion capture). An additional calculation of the slope of the linear line of best fit between average velocity 2 (timing lights) and average center of mass velocity (motion capture) was made to determine how the magnitude of both velocity measurements changed with respect to one another. For this slope calculation, the intercept was defined to be at zero, as zero velocity would be reported by both when no movement occurs. For interpretation, a strong positive correlation was defined as r ≥ 0.7, a moderate positive correlation was defined as 0.7 > r ≥ 0.3 and a weak positive correlation was defined as 0 < r < 0.3. A weak negative correlation was defined as 0 > r > -0.3, a moderate negative correlation was defined as -0.3 ≥ r > -0.7 and a strong negative correlation was defined as r ≤ -0.7.
Correlations were made intra-individually for each trial. We selected intra-individual correlations as they are a widely used analysis approach for biomechanics research that seeks to evaluate trends in biomechanical variables across a range of movement conditions (Honert et al. 2022; Matijevich et al. 2019; Savelberg and Lange 1999). All correlation, averages and standard deviations were calculated in Excel.
Results
Trials conducted in this study encompassed a wide range of forward accelerations (Table 1) and forward velocities (1.52-5.38 m/s). On average, none of the changes in velocity measured with timing lights had a strong positive correlation with the relative AP impulse (Table 2).
Table 1: AP impulse range for each participant. Ten rows represent the 10 participants. AP impulse range is reported as minimum AP impulse - maximum AP impulse for a participant.
Participant | AP impulse range (min-max) |
|---|---|
1 | 0.22-0.78 |
2 | 0.36-0.88 |
3 | 0.12-0.73 |
4 | 0.19-0.90 |
5 | 0.13-0.94 |
6 | 0.04-0.99 |
7 | 0.20-0.98 |
8 | 0.24-0.71 |
9 | 0.11-0.81 |
10 | 0.19-0.79 |
Table 2: Correlation coefficient (r) between the relative AP impulse and change in velocity calculated by timing lights data. Ten rows represent the 10 participants. Three correlation coefficients are determined by comparing each change in velocity as calculated by timing lights. Average correlation coefficient is reported as average ± STD.
Participant | r | r | r |
|---|---|---|---|
| Vavg,2 - Vavg,1 | Vavg,3 - Vavg,2 | Vavg,3 - Vavg,1 |
1 | -0.04 | 0.15 | 0.12 |
2 | -0.64 | 0.62 | -0.02 |
3 | -0.34 | 0.47 | 0.23 |
4 | -0.08 | 0.43 | 0.43 |
5 | 0.00 | 0.16 | 0.19 |
6 | -0.01 | 0.26 | 0.24 |
7 | -0.34 | 0.19 | -0.20 |
8 | -0.54 | 0.53 | -0.02 |
9 | 0.75 | -0.49 | 0.22 |
10 | -0.01 | 0.40 | 0.31 |
Average r | -0.01 ± 0.39 | 0.27 ± 0.31 | 0.15 ± 0.18 |
r - correlation coefficient, STD - standard deviation, Vavg - average velocity
The trials for a single participant are shown on a correlation plot to visualize the relationship between the change in velocity measurements with timing lights and the relative AP impulse (Figure 3). Comparison between forward acceleration state estimated by motion capture data and the relative AP impulse revealed a strong correlation (Table 3).
Figure 3. Velocity change calculated from timing lights (Vavg,3 - Vavg,2) vs relative AP impulse ( N=1). Each dot represents a single trial, with the x-axis corresponding to change in velocity calculated from timing lights and the y-axis corresponding to the relative AP impulse measured from the force plate. The black line represents the regression line. The correlation coefficient for participant 10 was determined to be 0.40, as shown in Table 2.
Table 3: Correlation coefficient (r) between the relative AP impulse and change in velocity calculated by motion capture data (velcoitytoe off – velocityfoot contact). Ten rows represent the 10 participants. Average correlation coefficient is reported as average ± STD.
Participants | r |
|---|---|
1 | 0.73 |
2 | 0.64 |
3 | 0.76 |
4 | 0.75 |
5 | 0.89 |
6 | 0.96 |
7 | 0.89 |
8 | 0.78 |
9 | 0.89 |
10 | 0.78 |
Average r | 0.81 ± 0.10 |
r - correlation coefficient, STD - standard deviation
The trials for a single participant are shown on a correlation plot to visualize the relationship between the change in velocity measurements with motion capture and the relative AP impulse (Figure 4). Furthermore, the comparison of velocity values estimated via motion capture (average center of mass velocity) with timing lights (average velocity 2) showed a strong positive correlation (Table 4).
Figure 4: Velocity change calculated from motion capture (V_toe off - W_heel strike) vs relative AP impulse (N=1). Each dot represents a single trial, with the x-axis corresponding to change in velocity calculated from motion capture and the y-axis corresponding to the relative AP impulse measured from the force plate. The black line represents the regression line. The correlation coefficient for participant 10 was determined to be 0.78, as shown in Table 3.
Table 4: Correlation coefficient between velocity values measured by timing lights (average velocity 2) and motion capture for each participant. Averages are reported as average ± STD. The slope is reported as the slope of the line of best fit for each participant’s data set, with the intercept set to zero.
Participant | r | Slope |
|---|---|---|
1 | 0.95 | 0.998 |
2 | 0.68 | 1.083 |
3 | 0.62 | 0.942 |
4 | 0.89 | 1.006 |
5 | 0.78 | 1.079 |
6 | 0.87 | 0.976 |
7 | 0.70 | 0.941 |
8 | 0.55 | 0.991 |
9 | 0.97 | 0.945 |
10 | 0.83 | 1.041 |
Average | 0.78 ± 0.14 | 1.000 ± 0.053 |
r - correlation coefficient, STD - standard deviation
The trials for a single participant are shown on a correlation plot to visualize the relationship between velocity measures with timing lights and motion capture (Figure 5).
Figure 5. Velocity measured by timing lights (average velocity2) vs velocity measured by motion capture for an example participant (N=1). Each dot represents a single trial, with the x-axis corresponding to average velocity 2 measured by timing lights and th ey-axis corresponding to the average velocity measured by motion capture. The black line represents the unity line (y=x). The correlation coefficient for participant one was determined to be 0.95, as shown in Table 4.
On average, the correlation between the relative AP impulse and average velocity 2 - average velocity 1 measured with timing lights was weak (r = -0.01). Four participants showed a weak negative correlation, four a moderate negative correlation, one a strong positive correlation and one no correlation (Table 2).
On average, the correlation between the relative AP impulse and average velocity 3 - average velocity 2 measured with timing lights was weak (r = 0.27). One participant showed a moderate negative correlation, four a weak positive correlation and five a moderate positive correlation (Table 2).
On average, the correlation between the relative AP impulse and average velocity 3 - average velocity 1 measured with timing lights was weak (r = 0.15). Three participants showed a weak negative correlation, five a weak positive correlation and two a moderate positive correlation (Table 2).
On average, the correlation between the relative AP impulse and velocity toe off – velocity foot contact measured with motion capture was strong (r=0.81). One participant showed a moderate positive correlation and nine a strong positive correlation (Table 3).
On average, the correlation between average velocity 2 (timing lights) and average center of mass velocity (motion capture) was strong (r = 0.78). Three participants showed a moderate positive correlation, and seven showed a strong positive correlation. The average slope of the line of best fit was determined to be 1.000 with a standard deviation of 0.053, indicating very little variability in the data (Table 4).
Discussion
The purpose of this study was to determine if there are alternative methods for measuring forward acceleration during overground running studies when force plates are not available, and the gold standard relative AP ground reaction force impulse cannot be calculated. Two potential alternative methods explored to estimate forward acceleration were the change in velocity measured with timing lights and the change in velocity measured with motion capture. We found a weak negative to weak positive correlation between change in velocity measured with timing lights and the relative AP impulse measured with force plates. Therefore, the hypothesis that change in velocity measured with timing lights would have a strong positive correlation with force plate measures (relative AP impulse) is rejected (H1). There was a strong positive correlation between change in velocity measured with motion capture and the relative AP impulse measured with force plates. Therefore, the hypothesis that change in velocity measured with motion capture would have a strong positive correlation with force plate measures (relative AP impulse) is accepted (H2). Results of this study support the debate on if and how forward acceleration should be controlled during locomotion studies.
The experimental protocol was designed with the aim of obtaining a wide range of relative AP impulses (minimum of 0.04; maximum of 0.99, Table 1). This was done to ensure that there would be a wide variety of forward acceleration states to compare across methods. This was successfully achieved by verbally instructing participants to either run at a steady-state (remain at a constant velocity), accelerate (increase velocity) or decelerate (decrease velocity) over the measurement distance.
Change in velocity measured with timing lights is not a reliable alternative to relative AP impulse measured with force plates when determining the forward acceleration of participants. For some participants, the correlation between relative AP impulse and timing lights was negative, demonstrating change in velocity measured with timing lights could yield the opposite of the expected results, misguiding researchers regarding the forward acceleration. Furthermore, there was a large variance in the data across participants (Table 2). One possible explanation of these results is an inherent limitation of the timing lights wherein a certain part of the body crossing a set of timing lights may trigger their sensors. As such, the timing light measurements may not always correspond to the center of mass motion of the participants. For example, for a pair of timing lights, the participant’s hand may trigger the sensor of the first timing light, while the participant’s torso may trigger the second timing light. This nuance may explain why timing lights are an unreliable indicator of center of mass change in velocity and yield a low correlation with the relative AP impulse. While the timing light manufacturer reportedly accounts for this effect, it is unknown if other body parts crossing the timing lights could still confound results. Another possible explanation is our timing lights were set up spatially before, over and after the force plate. While the relative AP impulse is only calculated over a stance phase from foot contact to toe off, change in velocity between two sets of timing lights covers a larger time window that may not correspond exactly to the stance phase of interest. Rather than covering a single stance phase, a single pair of timing lights covered a distance of 0.9m. As the relative AP impulse only considers a single stance phase, while timing light measurements are made independently of the number of stance phases, these approaches varied slightly in their data collection area. Since these two approaches use measurements from different locations/times, it is unsurprising that we did not find a strong correlation. Therefore, our results indicate that using timing lights to calculate change in velocity in place of the relative AP impulse is not a method that should be considered.
Change in velocity measured with motion capture is a feasible alternative to relative AP impulse measured with force plates when determining the forward acceleration of participants. When comparing relative AP impulse with change in velocity measured by motion capture (Equation 1), the average correlation was strong and positive (r = 0.81). An advantage of the motion capture method is that it includes measurements from a time window equivalent to the AP impulse method, which is the time window from foot contact to toe off. The standard deviation of the correlation coefficient was 0.10, indicating low variability across participants (Table 3). These results indicate that the changes in center of mass velocity approximated by pelvis motion capture markers are an accurate indicator of the net forward acceleration. These experimental results suggest changes in approximate center of mass velocity over the stance phase of running are a reliable biomechanical signal for estimating acceleration state. While we estimated center of mass acceleration using a cluster of pelvis motion capture markers and optical motion capture in this study, center of mass velocity can also be measured with other measurement approaches, such as by tracking a single sacral optical motion capture marker (Napier et al. 2020), signal integration of IMUs placed near the center of mass (Cardarelli et al. 2020), a complementary filter method with multiple IMUs on the body (Mohamed Refai et al. 2020), machine learning applied to sacral acceleration signals (Alcantara et al. 2021), etc. Identifying a signal of interest, for example the center of mass velocity, flexibly empowers researchers to then select a measurement tool that best aligns with their experimental protocol and study constraints. Because all measurement systems, such as wearables, for example, have error and limitations, there are great advantages to first leveraging high accuracy tools like optical motion capture to identify the signal(s) of interest (Matijevich et al. 2020). While optical motion capture was used here, it is possible that other more portable or wearable forms of motion capture (e.g., IMU-based, markerless.) may also be a viable solution for quantifying forward acceleration. Therefore, our results indicate that calculating change in velocity using motion capture data can be used in place of the relative AP impulse measured by force plates. However, there is room for further debate regarding the feasibility and ease of both of these methods, and whether one method should be considered the “gold-standard”.
A secondary purpose of this study was to compare velocity measured with timing lights and motion capture to determine if these methods can be used interchangeably. We found a strong positive correlation between change in velocity measured with timing lights and motion capture. The hypothesis that timing lights and motion capture can be used interchangeably when measuring velocity is accepted (H3).
When comparing velocity measurements made by timing lights (average velocity 2) and motion capture (average center of mass velocity), the average correlation was strong and positive (r = 0.78). The standard deviation of the correlation coefficient for this comparison was 0.13, indicating little variability between participants (Table 4). For this comparison, we elected to use average velocity 2 because this is the velocity measurement that is most similar to the time and location of velocity measured with motion capture (over the force plate). This may be a possible reason why timing lights show a high correlation when measuring velocity but a minimal correlation to AP impulse. As the AP impulse comparison was made with multiple sets of timing lights, the measurement distance for that comparison was greater than just the force plate, which is where the AP impulse measurements are calculated. However, for the velocity comparison, motion capture and average velocity 2 are collected over the same distance, which may explain the high correlation. The average slope of the linear line of best fit was 1.00 ± 0.053, indicating that the velocity values determined by each method change linearly with one another in a one-to-one ratio. An example participant shown in Figure 3 shows this trend, as there is a linear relationship between the two velocity measurements. A potential reason for a correlation that was not exactly one could be differences in the tracking period. Motion capture measured velocity for the stance phase whereas the timing lights were at fixed locations spanning the force plate. Across multiple trials, participants may have contacted the force plate at slightly different locations.
Recommendations for researchers moving forward
These findings provide researchers with alternative methods for quantifying forward acceleration during overground running studies in instances when force plates are not available. This may occur when researchers do not have a force plate or when researchers want to conduct studies outside the lab). Namely, the change in velocity of motion capture markers that approximate the location of the center of mass can be used to accurately determine the forward acceleration state. The findings of this study also give researchers confidence in using timing lights and motion capture interchangeably for measuring forward velocity. Broadly, we encourage biomechanics researchers to measure and control for forward acceleration during overground locomotion studies and to report the methods used for quantifying forward acceleration state.
Limitations and future research opportunities
The experimental design used in this study had limitations, indicating opportunities for future research. One limitation was the absence of walking movements in the study. Therefore, the reliability of using these methods for quantifying the forward acceleration of walking is unknown. A second limitation of this study is that pelvis markers were assumed to approximate the location of the participant’s center of mass. This approximation does not account for differences in body types as the location of center of mass may vary across individuals. Another limitation is there was no exploration of the reliability of using alternative motion capture marker sets, such as a single sacrum motion capture marker to approximate center of mass velocity (Bruening et al. 2018) or whole-body tracking with modeling to obtain the center of mass). Furthermore, this study was conducted in a laboratory setting with a small sample size (10 participants). Future research should be conducted outside the lab with a larger sample size to determine if the results of this study are applicable to research conducted in non-laboratory settings. An additional limitation of this work was that we used the relative AP impulse, rather than the absolute net AP impulse magnitude, as an indicator of the direction and magnitude of running acceleration. While this method was used to allow for comparison across a range of self- selected running speeds, future studies may explore the association between acceleration methods at a fixed running speed where net AP impulse can be more readily compared. Another limitation of this research is that we only used Pearson corre-lation coefficients to find the association between measurement methods. Other relative and absolute accuracy metrics (e.g., mean absolute percent error, root mean square error, Bland-Altman plots) could be used in the future to quantify the agreement between methods.
Finally, as mentioned previously, the timing lights were set up spatially before, over and after the force plate. Since the relative AP impulse measurement was only made over the force plate distance, this variability in measurement distance may have affected the results. We encourage researchers to explore the use of timing lights to measure acceleration in more depth for future studies by manipulating the timing light organization. Future researchers could have two sets of timing lights over the force plate distance to increase the similarity of measurement location.
Conclusion
This evaluation of overground running acceleration was intended to support the debate on if and how forward acceleration should be controlled for during locomotion studies. The primary purpose of this study was to identify if measuring change in velocity with two different sets of equipment, timing lights and motion capture, are accurate alternatives to relative AP impulse, which is measured by force plate, when quantifying overground running acceleration. We found that changes in velocity measured with timing lights are not a reliable surrogate for relative AP impulse. However, change in velocity measured with motion capture can be used as a substitute to relative AP impulse. A secondary purpose of this study was to determine if velocity measurements made from motion capture and timing lights can be used interchangeably. The results of this study indicate that both methods can reliably be used interchangeably. The findings of this study provide future researchers with a reliable method to quantify forward acceleration when force plates are not available. With this method, researchers can reliably control for trials that are or are not steady-state locomotion, giving them confidence in the interpretation of kinetics and kinematics variables across different trials and participants.
References
Alcantara, R. S., Day, E. M., Hahn, M. E. and Grabowski, A. M. (2021). ‘Sacral acceleration can predict whole-body kinetics and stride kinematics across running speeds. PeerJ, 9, 11199-11199.
Breine, B., Malcolm, P., Frederick, E. C. and Declercq, D. (2013). ‘Initial foot contact patterns during steady state shod running. Footwear Science, 5(sup1), 81-82. Brown, T. N., O’donovan, M., Hasselquist, L., Corner, B. and Schiffman, J. (2014). ‘Body borne loads impact walk-to-run and running biomechanics. Gait and Posture, 40(1), 237-242.
Bruening, D. A., Pohl, M. B., Takahashi, K. Z. and Barrios, J. A. (2018). ‘Midtarsal locking, the windlass mechanism, and running strike pattern: A kinematic and kinetic assessment. Journal of Biomechanics, 73, 185-191.
Caekenberghe, V., Segers, V., Aerts, P., Willems, P., Clercq, D. and D. (2013). Joint kinematics and kinetics of overground accelerated running versus running on an accelerated treadmill. Journal of the Royal Society Interface, 10(84), 20130222-20130222.
Cardarelli, S., Mengarelli, A., Tigrini, A., Strazza, A., Nardo, F. D., Fioretti, S. and Verdini, F. (2020). Single IMU Displacement and Orientation Estimation of Human Center of Mass: A Magnetometer-Free Approach. IEEE Transactions on Instrumentation and Measurement, 69(8), 5629-5639.
Day, E. M., Alcantara, R. S., Mcgeehan, M. A., Grabowski, A. M. and Hahn, M. E. (2021). Low-pass filter cutoff frequency affects sacralmounted inertial measurement unit estimations of peak vertical ground reaction force and contact time during treadmill running. Journal of Biomechanics, 119, 110323-110323.
Gill, N., O’leary, T., Roberts, A., Liu, A., Roerdink, M., Greeves, J. and Jones, R. (2023). Enforcing walking speed and step-length affects joint kinematics and kinetics in male and female healthy adults. Gait and Posture, 103, 223-228.
Havas, E., Lehtonen, M., Vuorela, J., Parviainen, T. and Vihko, V. (2000). Albumin clearance from human skeletal muscle during prolonged steady-state running. Experimental Physiology, 85(6), 863-868.
Honert, E. C., Hoitz, F., Blades, S., Nigg, S. R. and Nigg, B. M. (2022). Estimating Running Ground Reaction Forces from Plantar Pressure during Graded Running. Sensors, 22(9), 3338-3338.
Matijevich, E. S., Branscombe, L. M., Scott, L. R. and Zelik, K. E. (2019). Ground reaction force metrics are not strongly correlated with tibial bone load when running across speed and slopes: Implications for science, sport and wearable tech. PloS One, 14(1), 210000-0210000.
Matijevich, E. S., Scott, L. R., Volgyesi, P., Derry, K. H. and Zelik, K. E. (2020). Combining wearable sensor signals, machine learning and biomechanics to estimate tibial bone force and damage during running. Human Movement Science, 74, 102690-102690.
Napier, C., Jiang, X., Maclean, C. L., Menon, C. and Hunt, M. A. (2020). The use of a single sacral marker method to approximate the centre of mass trajectory during treadmill running. Journal of Biomechanics, 108, 109886-109886.
Refai, M., Beijnum, M. I. V., Buurke, B.-J. F., Veltink, J. H. and H, P. (2020). Portable Gait Lab: Instantaneous centre of mass velocity using three inertial measurement units. 2020 IEEE Sensors.
Riley, P. O., Dicharry, J., Franz, J., Croce, U. D., Wilder, R. P. and Kerrigan, D. C. (2008). ‘A Kinematics and Kinetic Comparison of Overground and Treadmill Running. Medicine and Science in Sports and Exercise, 40(6), 1093-1100.
Rodman, C. H., and Martin, A. E. (2020). Quantification of spatiotemporal parameter behavior during walking speed transitions. Journal of Biomechanics, 112, 110068-110068. Ruder, J., Tenforde, S. T., Mulloy, A., Davis, F. and S, I. (2019). Relationship of Foot Strike Pattern and Landing Impacts during a Marathon. Medicine and Science in Sports and Exercise, 51(10), 2073-2079.
Savelberg, H. H. C. M., and Lange, A. L. H. (1999). Assessment of the horizontal, fore-aft component of the ground reaction force from insole pressure patterns by using artificial neural networks. Clinical Biomechanics (Bristol), 14(8), 585-592.
Vanrenterghem, J., Gormley, D., Robinson, M. and Lees, A. (2010). ‘Solutions for representing the whole-body centre of mass in side cutting manoeuvres based on data that is typically available for lower limb kinematics. Gait and Posture, 31(4), 517-521.
White, J. A., Pomfret, D., Rennie, S., Gissane, C., Wong, J. and Ford, M. (1998). ‘Fluid replacement needs of well-trained male and female athletes during indoor and outdoor steady state running. Journal of Science and Medicine in Sport, 1(3), 131-142.
Williams, C., Powell, J. H. and W, D. (2017). ‘Lower Extremity Joint Work During Acceleration, Deceleration, and Steady State Running. Journal of Applied Biomechanics, 33(1), 56-63.
Willwacher, S., Oberländer, K. D., Mai, P., Mählich, D., Kurz, M., Koopmann, T., Fohrmann, D., Kantarev, A. and Kersting, U. G. (2021). A new method for measuring treadmill belt velocity fluctuations: effects of treadmill type, body mass and locomotion speed. Scientific Reports,11(1), 2244-2244.
Yang, F., and King, G. A. (2016). ‘Dynamic gait stability of treadmill versus overground walking in young adults. Journal of Electromyography and Kinesiology, 31, 81-87.