Biomechanical Signatures of Head Impacts Across Men's and Women's Contact Sports

Introduction: : Concussion and repetitive head impacts remain a significant concern in contact sports, particularly among youth and collegiate athletes, with growing evidence suggesting that both concussive and subconcussive impacts may contribute to long-term neurological changes. Wearable sensors, such as instrumented mouthguards (iMGs), offer on-field biomechanical data to inform brain injury risk, typically reporting peak linear and angular acceleration metrics. However, most studies only investigate single sports, and substantial variability in sensor specifications and data processing pipelines across studies has limited meaningful cross-sport comparisons. In addition, most existing literature emphasizes peak kinematics, with limited consideration of other important biomechanical factors such as impulse duration, directional loading, and frequency content, all of which may affect brain injury risk. Differences in equipment (e.g., helmets), gameplay dynamics, and sex may further influence head impact biomechanics between sports, complicating direct comparisons.

To address these challenges, we curated and uniformly processed a large multisport iMG dataset containing verified direct head impacts from men’s American football, men’s ice hockey, women’s rugby, and women’s soccer. This standardized approach enabled a controlled examination of cross-sport biomechanical differences using features associated with brain injury risk. Our two primary objectives were: (i) to compare resultant and directional kinematic features, impulse durations, and frequency characteristics across sports, and (ii) to determine whether these impacts cluster by sport using unsupervised learning techniques. Through this work, we aim to identify shared and sport-specific patterns in head impact biomechanics that could inform injury mitigation strategies across contact sports.

Materials and

Methods: : We curated a high-confidence dataset of video-verified, direct head impacts from university-level men’s American football, men’s ice hockey, women’s rugby, and women’s soccer. Impacts were collected using three different instrumented mouthguards (iMGs), and a uniform processing pipeline was applied to standardize sampling frequency, impact duration window, and filtering. All data were uniformly sampled and truncated to match a 1000 Hz, 50 ms window (10 ms pre- and 40 ms post-trigger). A 4th-order Butterworth filter (300 Hz for linear, 180 Hz for angular) was applied and angular velocities were differentiated to determine angular accelerations. Biomechanical features extracted included directional and resultant peak kinematics, impulse durations, 3D acceleration vectors, sagittal/coronal coupling ratios, and wavelet-derived frequency features (peak frequency, bandwidth). For objective i) we compared features across sports using Kruskal-Wallis tests with Bonferroni-corrected post-hoc analysis. For clustering in objective ii) we applied k-means algorithm to impact magnitude and frequency features and trained a t-distributed stochastic neighbor embedding (t-SNE) model to determine if sport-specific clusters arise.

Results, Conclusions, and Discussions:: Our final database included 1,250 video-verified direct head impacts across four sports: men’s hockey (n = 146), men’s football (n = 232), women’s rugby (n = 306), and women’s soccer (n = 566). Significant cross-sport differences were observed in peak kinematics (Fig.1), impulse durations, and directionality (Fig.2). Men’s football exhibited the highest resultant median PLA, PAV, and PAA, while women’s soccer consistently had the lowest. However, directional analyses revealed women’s soccer to have unexpectedly high sagittal-plane kinematics. Women’s soccer demonstrated primarily upward linear impacts and sagittal-plane angular accelerations, while men’s football showed backward linear and coronal angular directions. Additionally, median impulse durations were longest in men’s hockey and women’s rugby.

K-means clustering identified k = 2 as the optimal number of clusters for both magnitude and frequency features, with impacts separating into lower and higher magnitude and frequency groups, respectively. Women’s soccer impacts were mostly low magnitude and low frequency, while men’s football and women’s rugby largely represented the high magnitude, low frequency group (Fig.3). t-SNE projections revealed tight clustering for women’s soccer, consistent with its repetitive heading mechanics, while other sports displayed overlapping patterns.
Our kinematic values align with recent iMG studies, falling on the lower end of previously reported ranges, likely due to improved sensor accuracy and standardized processing. Our dataset included both men’s and women’s data, however, preliminary observations showed comparable ranges of peak kinematics between sexes, challenging the theory that weaker neck muscles in women lead to more severe impacts. Additionally, helmeted impacts did not exhibit reduced durations or lower frequency content, stressing the need for further research into the effectiveness of helmets for impacts commonly sustained during play.

Overall, combining the findings from the two objectives, we observe that despite some notable distribution differences in biomechanical parameters, direct head impacts in the sports we examined have largely overlapping biomechanical characteristics. These findings highlight the value of standardized multisport comparisons in understanding injury risk and call for broader, harmonized datasets to inform future protective strategies across contact sports.

Acknowledgements and/or References (Optional)::