Abstract
his study investigates the relationship between total running experience, de- fined as cumulative years of running multiplied by weekly mileage, and variability in lower leg joint kinematics during treadmill running. Twenty-seven male athletes partici- pated, running while kinematic and kinetic data were collected. Linear regression revealed significant negative correlations between total running experience and variability in both knee and ankle joint range of motion (ROM). Specifically, ankle ROM variability (p= 0.001, R 2 = 0.35) and knee ROM variability (p= 0.002, R 2 = 0.32) were reduced in runners with more experience. A stepwise regression model further identified ankle ROM variability as a significant predictor (p= 0.033), explaining 44.25% of the variance in total running experience. A significant positive correlation between running experience and instanta- neous vertical loading rate (IVLR) (p= 0.025, R 2 = 0.15) suggests that more experienced runners generate higher load rates. These findings indicate that more experienced runners exhibit more consistent and stable movement patterns, reflecting refined motor control. The results support the hypothesis that greater running experience is associated with reduced variability in movement patterns within a controlled environment, providing insights into the mechanisms that could contribute to enhanced performance and injury prevention. Keywords:kinematic variability; running experience; performance 1. Introduction Running is widely recognized for its health benefits, yet it carries notable risks, par- ticularly for novice
control. The results support the hypothesis that greater running experience is associated with reduced variability in movement patterns within a controlled environment, providing insights into the mechanisms that could contribute to enhanced performance and injury prevention. Keywords:kinematic variability; running experience; performance 1. Introduction Running is widely recognized for its health benefits, yet it carries notable risks, par- ticularly for novice runners. Overuse injuries remain a prevalent concern, with reported incidence rates ranging from 19% to 79%, often attributed to factors such as weekly mileage and cumulative years of running [1–3]. Contrary to traditional beliefs, recent research suggests that running experience may not be a factor in injury risk for novice runners [4]. This finding has redirected attention toward alternative explanations for the observed variability in injury rates across experience levels. Advances in biomechanics highlight the role of motor control strategies, which adapt and evolve with experience [5]. These adaptations influence coordination variability, de- fined as the movement pattern differences between segments, which is gaining recognition as a critical aspect of running biomechanics [6,7]. While this study quantifies variability using the standard deviation of sagittal plane joint kinematics, alternative mathematical approaches exist to assess movement variability. Lyapunov exponents measure the diver- gence of small perturbations in movement trajectories, providing insights into stability and adaptability [8]. Approximate entropy and sample entropy quantify movement com- plexity and predictability, while fractal analysis methods such as Detrended Fluctuation Sports2025,13, 58 https://doi.org/10.3390/sports13020058
Sports2025,13, 58 2 of 11 Analysis (DFA) assess long-range correlations in movement patterns. Additionally, vector coding techniques have been widely used to evaluate segmental coordination variability, particularly in running biomechanics [9,10]. These approaches offer deeper insights into inter-segmental interactions, which could be valuable for future investigations into motor control adaptations in experienced runners. Although these approaches provide advanced perspectives on variability, standard deviation remains a useful metric for comparing variability across different experience levels in running biomechanics. Empirical evidence suggests that trained runners demonstrate reduced movement variability compared to novices, indicating that cumulative weekly mileage and years of running, collectively termed ‘total running’, may refine motor patterns and stabilize coordination [8]. Movement variability can be categorized into two distinct forms: (1) skill acquisition variability, where higher variability reflects inconsistent motor control during the learning process, and (2) functional variability, where controlled movement adjustments enable experienced runners to optimize performance and adaptability to environmental changes. This distinction has been studied in motor control research [11,12], with findings suggest- ing that reduced variability in experienced runners reflects enhanced coordination and efficiency, rather than motor inflexibility [12]. These forms of variability play a role in running biomechanics, where the transition from high to controlled variability marks the progression from learning to skilled running performance. This dual function of variability, as a marker of motor refinement and adaptability, is consistent with findings across other sports, emphasizing the necessity for focused investigations in running biomechanics [13]. While movement variability has been extensively studied in sports such as cycling, race walking, and swimming [14–16], findings in running remain inconsistent, revealing a gap in understanding how experience shapes running biomechanics. Bridging this gap is essential to resolving the mechanisms responsible for possible running-related injuries and performance [17,18]. In addition to sagittal joint angles at the knee and angle, ankle angle at initial contact and instantaneous vertical loading rate (IVLR) are also markers of skill adaptation and stability in running. The ankle angle at initial contact is critical for assessing shock attenuation during the stance phase, with novice runners often exhibiting less stable landing mechanics [19]. IVLR, a
In addition to sagittal joint angles at the knee and angle, ankle angle at initial contact and instantaneous vertical loading rate (IVLR) are also markers of skill adaptation and stability in running. The ankle angle at initial contact is critical for assessing shock attenuation during the stance phase, with novice runners often exhibiting less stable landing mechanics [19]. IVLR, a measure of force application during early stance, has been correlated with training exposure and neuromuscular control, offering insights into injury susceptibility and motor refinement [20]. Although variability has been shown to decrease with experience, it is unlikely that this reduction continues indefinitely in a linear fashion. Research on motor learning suggests that movement variability may reach a plateau beyond which additional experience does not significantly alter coordination patterns [21]. This concept is well documented in skill acquisition studies, where performance improvements stabilize over time as an optimal movement strategy emerges. Future studies should investigate whether running experience follows a similar asymptotic trajectory, in which variability reductions become negligible beyond a certain threshold of training exposure. This current study examines the association between total running experience on lower extremity variability during controlled treadmill running. This study also includes exploratory metrics, such as . VO2 and IVLR, to examine whether physiological and me- chanical factors also influence kinematic variability in runners [19,20]. We hypothesize that greater running experience will be associated with reduced knee and ankle variabil- ity, reflecting a more stable and refined motor control strategy. Given that the knee and ankle play primary roles in impact absorption and propulsion, they are more sensitive to training effects than the hip [9,22]. While the hip contributes to global movement control, prior research suggests that distal joint variability is more directly influenced by long-term running exposure, making the knee and ankle more appropriate targets for analysis.
Sports2025,13, 58 3 of 11 Research has shown that segmental coordination patterns evolve with experience, particularly at the shank and foot, where greater adaptability emerges over time [9]. How- ever, at the joint level, running experience is typically associated with reduced movement variability, as motor refinement minimizes unnecessary motion [22]. By focusing on joint kinematics, rather than inter-segmental coordination, this study aims to evaluate how running experience is associated with motor control strategies at individual joints. Further, the inclusion of ankle angle at initial contact as a stance-phase-specific metric aligns with its role in assessing sagittal plane stability and control during running. As a secondary focus, this study investigates metabolic factors, specifically sub- maximal . VO2 , to explore their potential relationship with movement variability in runners. Sub-maximal . VO2 measured during steady-state running, serves as a proxy for running economy and reflects the efficiency of neuromuscular and metabolic systems under con- trolled conditions. Previous research has shown that improved running economy is associ- ated with refined motor patterns and enhanced performance, suggesting that sub-maximal . VO2may play a role in reducing variability during repetitive tasks [23,24]. 2. Materials and Methods A total of 27 male athletes were recruited for this study. All participants had prior experience running on the laboratory treadmill and completed a questionnaire documenting recent injuries, weekly mileage, and cumulative years of running. Inclusion criteria required participants to report no history of chronic overuse injuries within the six months preceding the study. Participant demographics and training characteristics are summarized in Table. Written informed consent was obtained from all participants before inclusion, and the study protocol was approved by the Brigham Young University Institutional Review Board (IRB #IRB2021-302). Table 1.Subject characteristics. Subject Age (Years) Height (cm) Mass (kg) Weekly Mileage (Miles) Years Running (Years) Total Running (Miles) . VO2 (mL·min −1 ·kg −1 ) 1 49 177.8 68.2 70 31 2170 51.6 2 28 177.8 67.3 25 15 375 52.4 3 20 185.42 71.4 55 1 55 49.1 4 32 180.34 68.2 90 17 1530 45.6 5 23 188 79.5 50 10 500 50.6 6 21 170.18 65.9
Weekly Mileage (Miles) Years Running (Years) Total Running (Miles) . VO2 (mL·min −1 ·kg −1 ) 1 49 177.8 68.2 70 31 2170 51.6 2 28 177.8 67.3 25 15 375 52.4 3 20 185.42 71.4 55 1 55 49.1 4 32 180.34 68.2 90 17 1530 45.6 5 23 188 79.5 50 10 500 50.6 6 21 170.18 65.9 48 8 384 44.5 7 24 177.8 64.1 15 11 165 47.7 8 30 177.8 71.8 60 15 900 48.4 9 25 167.6 63.6 30 9 270 59.2 10 18 180.3 65.9 45 4 180 51.8 11 18 177.8 55.5 50 4 200 53.1 12 20 180.3 68.2 60 7 420 50.9 13 26 188 70.3 20 15 300 43.2 14 18 175.3 57.2 50 4 200 54.8 15 37 180.3 80.3 40 25 1000 46.1 16 32 177.8 79.5 10 20 200 52.4 17 30 182.8 68.0 30 15 450 46.7 18 42 175.26 63.1 95 25 2375 36.2 19 25 182.88 82.0 65 6 390 44.6 20 21 181 74.0 75 10 750 44.2 21 21 172.5 63.0 60 8 480 40.8 22 19 178 66.3 45 3 135 49.5 23 44 182 69.4 50 31 1550 46.1 24 21 179.5 58.6 14 6 84 55.3 25 42 183.5 88.5 55 29 1595 52.4 26 23 180.5 73.5 55 10 550 50.8 27 45 177.5 65.8 45 25 1125 49.9 27.92 (9.38) 179.18 (4.65) 69.22 (7.73) 48.41 (21.31) 13.48 (9.05) 708.32 (651.2) 48.81 (4.87)
Sports2025,13, 58 4 of 11 Three-dimensional kinematic data were collected using a 13-camera Vicon motion capture system (Vicon Motion Systems, Oxford, UK) sampling at 240 Hz, while kinetic data were recorded from a force-instrumented treadmill (Bertec, Columbus, OH, USA) sampling at 960 Hz. Prior to each participant’s data collection, the motion capture system was calibrated following the manufacturer’s guidelines to ensure optimal accuracy of marker tracking. Retro-reflective markers (n = 24) were placed on anatomical landmarks of the lower body, including the medial and lateral epicondyles, medial and lateral malleoli, the first metatarsal head, and a triad on the heel (Figure). To ensure consistency in marker placement, the same researcher applied the markers for all participants. Marker placement was verified by a second researcher to reduce potential error. All participants were distance runners with low body fat percentages, which facilitated the accurate identification of bony landmarks. Rigid marker clusters with four non-collinear reflective markers were secured to the lateral aspects of the thigh and shank segments for tracking. Pelvic kinematics were captured using markers positioned at the bilateral anterior superior iliac spines (ASISs) and posterior superior iliac spines (PSISs). To assess metabolic responses, participants wore a portable metabolic measurement system (Cosmed K5, Rome, Italy) following marker placement.Sports 2025,13, x FOR PEER REVIEW 4of 11 20 21 181 74.0 75 10 750 44.2 21 21 172.5 63.0 60 8 480 40.8 22 19 178 66.3 45 3 135 49.5 23 44 182 69.4 50 31 1550 46.1 24 21 179.5 58.6 14 6 84 55.3 25 42 183.5 88.5 55 29 1595 52.4 26 23 180.5 73.5 55 10 550 50.8 27 45 177.5 65.8 45 25 1125 49.9 27.92 (9.38) 179.18 (4.65) 69.22 (7.73) 48.41 (21.31) 13.48 (9.05) 708.32 (651.2) 48.81 (4.87) Three-dimensional kinematic data were collected using a 13-camera Vicon motion cap- ture system (Vicon Motion Systems, Oxford, UK) sampling at 240 Hz, while kinetic data were recorded from a force-instrumented treadmill (Bertec, Columbus, OH, USA) sampling at 960 Hz. Prior to each participant’s data collection, the motion capture system was cali- brated following the manufacturer’s guidelines to ensure
48.81 (4.87) Three-dimensional kinematic data were collected using a 13-camera Vicon motion cap- ture system (Vicon Motion Systems, Oxford, UK) sampling at 240 Hz, while kinetic data were recorded from a force-instrumented treadmill (Bertec, Columbus, OH, USA) sampling at 960 Hz. Prior to each participant’s data collection, the motion capture system was cali- brated following the manufacturer’s guidelines to ensure optimal accuracy of marker track- ing. Retro-reflective markers (n = 24) were placed on anatomical landmarks of the lower body, including the medial and lateral epicondyles, medial and lateral malleoli, the first metatarsal head, and a triad on the heel (Figure 1). To ensure consistency in marker place- ment, the same researcher applied the markers for all participants. Marker placement was verified by a second researcher to reduce potential error. All participants were distance run- ners with low body fat percentages, which facilitated the accurate identification of bony landmarks. Rigid marker clusters with four non-collinear reflective markers were secured to the lateral aspects of the thigh and shank segments for tracking. Pelvic kinematics were captured using markers positioned at the bilateral anterior superior iliac spines (ASISs) and posterior superior iliac spines (PSISs). To assess metabolic responses, participants wore a portable metabolic measurement system (Cosmed K5, Italy) following marker placement. Figure 1. Marker setup and placement for all subjects. Participants completed a 5 min warm-up at a constant speed of 3.83 m/s to acclimate to the setup and achieve steady-state V 6O 6levels. Following the warm-up, participants per- formed a 10 min treadmill running trial at the same speed. Kinematic and kinetic data were captured during the final 30 s of minutes 3, 5, 7, and 9, resulting in approximately 30 strides per collection interval. Metabolic data were continuously recorded throughout the trial and averaged over consecutive 10 s periods. Steady-state V 6O 6values were confirmed, with the Figure 1.Marker setup and placement for all subjects. Participants completed a 5 min warm-up at a constant speed of 3.83 m/s to acclimate to the setup and achieve steady-state . VO2 levels. Following the warm-up, participants performed a 10 min treadmill running trial
the trial and averaged over consecutive 10 s periods. Steady-state V 6O 6values were confirmed, with the Figure 1.Marker setup and placement for all subjects. Participants completed a 5 min warm-up at a constant speed of 3.83 m/s to acclimate to the setup and achieve steady-state . VO2 levels. Following the warm-up, participants performed a 10 min treadmill running trial at the same speed. Kinematic and kinetic data were captured during the final 30 s of minutes 3, 5, 7, and 9, resulting in approximately 30 strides per collection interval. Metabolic data were continuously recorded throughout the trial and averaged over consecutive 10 s periods. Steady-state . VO2 values were confirmed, with the final 3 min average used to represent the entire trial. Analyses focused on data collected after the initial 3 min to account for the stabilization of running biomechanics. Joint kinematic data were derived from three-dimensional marker positions captured during running trials. Marker positions were processed using a 20 Hz low-pass, second- order Butterworth filter in Visual3D (C-Motion, Germantown, MD, USA). Ankle angle at initial contact was calculated as the sagittal plane ankle joint angle at the moment of initial contact, defined as the first frame when vertical ground reaction force exceeded 30 N. This approach was automated within Visual3D software v2024.06.1, ensuring consistent identification across trials. Instantaneous vertical loading rate (IVLR) was calculated from
Sports2025,13, 58 5 of 11 the vertical ground reaction force (GRF) data to capture impact mechanics during the stance phase. IVLR was defined as the derivative of the vertical GRF just prior to the impact peak (IP). For participants without an identifiable IP—commonly observed in forefoot strike (FFS) patterns—IVLR was taken at 13% of stance [25]. Statistical Analysis All initial data visualizations and statistical analyses were performed using MATLAB (Version R2023b, 23.2, The MathWorks Inc., Natick, MA, USA) and R (Version 2024.04.2+764 (2024.04.2+764), R Core Team, 2024). Segment variability was quantified as the standard deviation of joint kinematic data over each 30 s collection period. Descriptive statistics are reported as means and standard deviations, and data normality was verified using the Shapiro–Wilk test. Relationships between total running experience and joint kinematic variability were assessed using linear regression analyses. To address potential false discovery rates, the Benjamani–Hochberg method [26] was applied to regression-derived p-values. This method controls the false discovery rate, which is the expected proportion of incorrect rejections among all rejected hypotheses, creating a balance between statistical power and error rate. Significant predictors were incorporated into the stepwise regression model, using the Akaike Information Criterion (AIC) to identify the final model [27]. The AIC is a model selection criterion that evaluates the relative quality of statistical models for a given dataset by balancing model fit with complexity, which helps avoid overfitting. Statistical significance was inferred fromp< 0.05. 3. Results 3.1. Correlation Analysis Prior to regression analyses, Pearson correlation coefficients were calculated to assess the relationships between total running experience and key variables. Significant negative correlations were observed between total running experience and knee ROM variability (r =−0.57,p= 0.002), ankle ROM variability (r =−0.59,p= 0.001), and ankle angle at initial contact variability (r =−0.49,p= 0.008). A significant positive correlation was found between total running experience and IVLR (r = 0.39,p= 0.025). 3.2. Linear Regression Analysis Before conducting regression analyses, we assessed the normality of key dependent variables using the Shapiro–Wilk test to confirm the appropriateness of parametric statistical methods. The results indicated that knee ROM variability (W = 0.973,p= 0.696), ankle
variability (r =−0.49,p= 0.008). A significant positive correlation was found between total running experience and IVLR (r = 0.39,p= 0.025). 3.2. Linear Regression Analysis Before conducting regression analyses, we assessed the normality of key dependent variables using the Shapiro–Wilk test to confirm the appropriateness of parametric statistical methods. The results indicated that knee ROM variability (W = 0.973,p= 0.696), ankle ROM variability (W = 0.953,p= 0.255), ankle angle at initial contact variability (W = 0.969, p= 0.712), and instantaneous vertical loading rate (IVLR) (W = 0.983,p= 0.925) all met the assumption of normality (p> 0.05). Given this, parametric linear regression analyses were deemed appropriate for further statistical testing. Initial linear regression analyses indicated significant negative correlations between total running experience and variability measures in the lower extremity joints. Specifically, significant relationships were found for knee range of motion (ROM) variability (β=−0.57, p= 0.002, R 2 = 0.32) (Figure), ankle ROM variability ( β=−0.59,p= 0.001, R 2 = 0.35) (Figure), and ankle angle at initial contact variability (β=−0.49,p= 0.008, R 2 = 0.24). Additionally, there was also a significant positive correlation between total running and instantaneous vertical loading rate (IVLR) (β= 0.39, R 2 = 0.15;p= 0.025), suggesting that increases in running experience lead to greater load rates.
Sports2025,13, 58 6 of 11Sports 2025,13, x FOR PEER REVIEW 6of 11 p= 0.002, R 2 = 0.32) (Figure 2), ankle ROM variability (β =−0.59, p= 0.001, R 2 = 0.35) (Figure 3), and ankle angle at initial contact variability (β =−0.49, p= 0.008, R 2 = 0.24). Additionally, there was also a significant positive correlation between total running and instantaneous vertical loading rate (IVLR) (β = 0.39, R 2 = 0.15; p= 0.025), suggesting that increases in run- ning experience lead to greater load rates. Figure 2. Regression comparing Knee ROM variability, defined as standard deviation across the col- lection, compared to total years spent running. Figure 3. Regression comparing ankle ROM variability, defined as standard deviation across the col- lection, compared to total years spent running. 3.3. Stepwise Regression Analysis A stepwise regression analysis was performed. The initial model included knee ROM variability, ankle angle at initial contact, and ankle ROM variability. During the stepwise process, ankle angle at initial contact was removed due to its higher p-value, resulting in a final model that included knee ROM variability and ankle ROM variability as predictors. In the final model, ankle ROM variability remained a significant predictor (p= 0.033), indicating that greater total running experience is associated with reduced ankle range of motion variability. Although knee ROM variability was not statistically significant at the conventional 0.05 level (p= 0.053), it approached significance, suggesting a trend where in- creased running experience may be linked to reduced knee range of motion variability. The overall model explained 44.25% of the variance in total running experience (adjusted R 2 = 0.396, p<0.001). Figure 2.Regression comparing Knee ROM variability, defined as standard deviation across the collection, compared to total years spent running.Sports 2025,13, x FOR PEER REVIEW 6of 11 p= 0.002, R 2 = 0.32) (Figure 2), ankle ROM variability (β =−0.59, p= 0.001, R 2 = 0.35) (Figure 3), and ankle angle at initial contact variability (β =−0.49, p= 0.008, R 2 = 0.24). Additionally, there was also a significant positive correlation between total running and instantaneous vertical loading rate (IVLR) (β = 0.39,
Description
The study examines how running experience affects lower leg joint variability in male athletes.