Sports Biomechanics最新文献

Portable data collection devices and machine learning (ML) have been combined in autonomous movement analysis models for resistance training (RT) movements. However, input features for these models were mostly extracted empirically and subsequent models demonstrated limited interpretability and generalisability to real-world settings. This study aimed to investigate the utility of interpretable and generalisable modelling techniques and several data-driven feature extraction (FE) methods. This was achieved by developing machine learning movement analysis models for the barbell back squat and deadlift using markerless motion capture. 61 participants performed submaximal and maximal repetitions of both RT movements. Movement data was collected using two Azure Kinect cameras. Joint and segment kinematic variables were calculated from the collected depth imaging, and input features were extracted using traditional, manual FE methods and novel data-driven techniques. Classifiers were developed for several predefined technical deviations for both movements. Many of the addressed technical deviations could be classified with good levels of accuracy (≥70%) while the remainder were poor (55%-60%). Additionally, data-driven FE techniques were comparable to previous, traditional FE methods. Interpretable and generalisable modelling techniques can be utilised to good effect for certain classification tasks while data-driven FE techniques did not provide a consistent advantage over traditional FE methods.

Biomechanical measurements of accidental ankle sprain injuries are rare but make important contributions to a more detailed understanding of the injury mechanism. In this case study, we present the kinematics and kinetics of a lateral ankle sprain of a female athlete performing handball-specific fake-and-cut manoeuvres. Three-dimensional kinematics and kinetics were recorded and six previously performed trials were used as reference. Plantarflexion, inversion, and internal rotation angles were substantially larger than the reference trials and peaked between 190 and 200 ms after initial ground contact. We observed a highly increased inversion and internal rotation moment. However, compared to the non-injury trials the data also revealed a reduction in the second dorsiflexion moment peak. Ground reaction forces were lower throughout the injury trial. Other parameters at initial ground contact including ankle and hip position, step length, and the traction coefficient indicate that a preparatory maladjustment occurred. This study adds valuable contributions to the understanding of lateral ankle sprains by building upon previously published reports and considering the shoe-surface interaction as an important factor for injury.

Trail runners have been reported to be more injury prone than road runners. Limited past studies have examined the difference in the running biomechanics between the two groups of runners. More importantly, the effect of surface inclination has not been fully investigated. Hence, this study examined the effect of surface inclination on running biomechanics in trail and road runners. Twenty trails and 20 road runners were recruited in this study. Trail runners appeared to be more experienced and had longer training distance per week (p < 0.001) compared to road runners. All participants ran at a self-selected pace on an instrumented treadmill in three inclination conditions (i.e., level, +10% uphill and -10% downhill) in a random order. Vertical average loading rate (VALR), vertical instantaneous loading rate (VILR) and footstrike angle (FSA) were measured using established methods. Trail runners experienced greater VILR (p = 0.039, Cohen's d = 2.9) with a greater FSA (p = 0.002, Cohen's d = 1.1) during downhill running than road runners. No significant differences in VALR, VILR and FSA were found between the two groups during level and uphill running. Our findings provide potential biomechanical rationale to explain a higher injury incidence among trail runners.

We examined the effects of barbell load and footwear type on hip, knee, and ankle joint torques during barbell squats in CrossFit athletes. Sixteen participants performed squats at 50%, 70%, and 90% of their 3-repetition maximum, wearing either weightlifting shoes or conventional footwear. Kinematic and kinetic data were captured using a three-dimensional analysis system, enabling the calculation of joint torques and trunk flexion angles at the instant of peak total support torque. Peak total support torque increased with load in both footwear conditions, with no interactions between footwear type and load observed for peak total support torque or individual joint contributions. Regardless of load, weightlifting shoes resulted in a lower hip torque contribution (p = 0.017) and a higher knee torque contribution (p = 0.024) compared to conventional shoes. As load increased, knee torque contribution decreased (p < 0.001), while ankle torque contribution increased (p = 0.025), accompanied by, on average, small increases in trunk flexion angle. Peak total support torque and peak trunk flexion did not coincide, potentially explaining the modest effects of load and footwear type observed. Analysis throughout the entire squat cycle is needed to fully understand joint contributions, aiding in optimizing squat techniques and footwear choices.

The purpose of this study was to propose the TJ-index, which uses deflection distance and contact time as variables, as an index of somersault jumps in trampoline competitions, and to clarify the characteristics of the jumping motion of trampoline gymnasts as a case study. Ten participants, ranging from elite trampoline gymnasts to those at the national competition level, performed 10 consecutive backward somersaults and forward somersault half-twists in alternating sequence. A strong positive correlation was observed between the TJ-index and time of flight (r = 0.98, p < 0.01). The relationship between deflection distance and contact time, which together constitute the TJ-index, was classified into six types using Ward's cluster analysis method. Moreover, normalising deflection distance by body mass revealed individual jump characteristics. The bed's deflection distance may primarily result from either the gymnast's body mass or their active jumping motion. A strong negative correlation was observed between deflection distance per unit of body mass and contact time (r = -0.87, p < 0.01). These findings suggested the existence of body mass optimisation in trampoline gymnasts. The use of the TJ-index appears to have potential for identifying and classifying the jumping motion tendencies of trampoline gymnasts by type.

Modern inertial measurement units house multiple accelerometers with differing specifications. Previous studies have used both high-g and low-g accelerometers to measure peak tibial acceleration (TA) during running. This dual approach may introduce inconsistencies with data processing due to the different specifications. The purpose of this study was to evaluate the agreement between the high-g and low-g accelerometers in measuring peak axial and resultant TA across a range of running velocities. One hundred recreational runners ran on an athletics track at five self-selected velocities, ranging from very slow to very fast based on their comfortable training pace. Results indicated the difference in agreement between the high-g and low-g accelerometers was curvilinear, with bias shifting towards the high-g accelerometer at higher TA magnitudes. However, the predicted difference was small, only exceeding ±1 g when measuring axial TA approaching its maximum range (16 g). Based on the comparable performance of the high-g and low-g accelerometers, and the relatively high frequency of data clipping in the low-g accelerometer (209 of 495 trials; 42%), it is recommended to use the high-g accelerometer exclusively for measuring TA during running, simplifying data collection and processing requirements.

This study evaluates the ability of body segment kinematic data to identify skating tasks in ice hockey using machine learning models and compares the performance of models trained on different body segments. We employed XGBoost, Support Vector Machine and Random Forest models to classify four primary ice-hockey skating tasks: forward skating start and strides, skating stop & go, and skating into a wrist shot. Trunk, pelvis, thigh, shank, and foot segment centre of mass linear accelerations were derived from retro-reflective markers and used as inputs for feature engineering. The models were trained and evaluated using a 10-fold cross-validation stratified by participant. Overall, the machine learning models demonstrated strong performance, with mean accuracy scores ranging from 86.5% to 98.9%. The pelvis yielded the best overall performance, followed by the trunk and foot, whereas the thigh segment generally exhibited lower accuracies across models. These results indicate that prediction performance depends on the body segment kinematic data used as input. This study highlights the potential of body segment kinematic data for automated identification of ice hockey skating tasks, providing insights into sports analytics and player performance assessment.

This study aimed to assess whether training to increase knee flexion during the tennis serve improves performance and to explore the associated biomechanical changes across the body. Twenty junior tennis players were randomly allocated into control (standard in-season training) and training groups (received training to increase knee flexion during serve). Inertial sensors tracked full body and racket kinematics during five serves performed in pre- and post-training assessments. Racket velocity, impact height, and lower- and upper-body kinematics were compared. Training increased serve knee flexion by 31° (p < 0.001), leading to a 1.38 km/h increase racket velocity (p = 0.036) without affecting impact height (p = 0.331). Additionally, training increased: range of front leg knee extension (MD = 23.46°, p < 0.001) and extension velocity (MD = 54.28°/s, p = 0.008), hip range of motion (front: MD = 53.60°/s, p = 0.003; back: MD = 57.28°/s, p = 0.015), pelvis upward velocity (MD = 0.27 m/s, p < 0.001), and trunk contralateral flexion velocity (MD = 23.18°/s, p = 0.025). No main effects were found for shoulder internal rotation (p = 0.304) and elbow extension (p = 0.214) velocities. No changes were observed in the control group other than a decreased trunk contralateral flexion velocity (MD = -28.98°/s, p = 0.007). Specific training can, therefore, increase serve knee flexion. This study highlights that specific training to increase knee flexion can enhance serve performance by increasing racket velocity, without increasing upper limb joint contribution.

The undersomersault (Felge) is one of the key family of skills on the parallel bars in men's artistic gymnastics. At the highest level of competition, two distinct techniques, termed here 'deep pike' and 'hips close', are used to perform the skill. The aim of the present study was to determine the relative performance benefits of each technique in order to provide coaches with information to facilitate technique selection and gymnast preparation. A combination of kinematic analysis of Olympic performances and technique optimisation using computer simulation modelling was used to address this aim. The kinematic analysis found that both techniques had similar performance outcomes in terms of generating horizontal and vertical velocity at release, confirming that both techniques were fit for purpose. Results from the optimisation study found that the 'deep pike' had an advantage in generating vertical velocity, due to more time to perform work, and being more forgiving in terms of generating a larger release window for acceptable performance, whilst the 'hips close' technique was associated with requiring less effort (sum of joint torques squared). As the 'deep pike' requires more effort and a larger range of hip flexibility, choosing this technique will have implications for gymnast preparation.

Traditional sidestepping experiment often relies on simplified visual stimuli that lack ecological validity. This study aimed to develop a fully immersive, football-specific virtual reality (VR) system to examine knee biomechanics during sidestepping in response to realistic visual stimuli. Twelve male collegiate footballers performed unanticipated sidestepping in response to a virtual footballer avatar executing either non-deceptive (VF-ND) or deceptive (VF-D) dribbling. Despite similar approach velocity and stance time, participants exhibited greater knee flexion angles and abduction and internal rotation moments in VF-D trials. Secondary analyses compared the results with previously reported arrow-preplanned (A-PP) and arrow-unplanned (A-UP) trials from the same participants. Approaching velocity decreased, and stance time increased in the following order: A-PP, VF-ND, VF-D and A-UP. Knee flexion angles and abduction moments increased, while peak internal rotation moments decreased in the same order. These findings suggested that the VR-based approach imposed more realistic visuospatial and temporal constraints than traditional methods, enhancing ecological validity. As in real environment, players in VR can perceive subtle cues, distinguish deceptive from non-deceptive actions and adapt their movements accordingly. Practitioners should design their protocols to resemble real-world scenarios as closely as possible and interpret the biomechanical outcomes cautiously when comparing across different visual stimuli.