Identifying Biomechanical Risk Factors for Lower Limb Injuries in High Jump Athletes Using Penalised Logistic Regression

Abstract

Background. High jump athletes are exposed to considerably lower limb injury risk due to repetitive high-impact loading and asymmetrical force application during the approach and take-off phases. Despite the biomechanical demands of the event, limited research has examined the predictive value of combined kinematic and neuromuscular factors in identifying athletes at elevated risk of musculoskeletal injury. Objectives. This study aimed to identify biomechanical predictors of lower limb injury risk in competitive male high jump athletes using penalised logistic regression. Materials and Methods. Twenty-one male national-level high jump athletes (age 21.14 ± 2.22 years; height 187.04 ± 5.36 cm; body mass 74.09 ± 5.04 kg) underwent 3D motion capture, ground reaction force analysis, and surface electromyography. Key predictors included cadence (steps/min), pelvic obliquity (°), pelvic rotation (°), and muscle activation asymmetry (% difference in EMG amplitude between limbs). Injury classification followed the International Olympic Committee’s consensus criteria, with injury history verified by medical records. Correlation analyses were followed by LASSO logistic regression with leave-one-out cross-validation. Model performance was assessed using AUC, sensitivity, specificity, predictive values, F1 score, calibration slope, intercept, and Brier score. Results. Four variables were retained in the final model: cadence (OR = 1.60, p = 0.021), pelvic obliquity (OR = 1.48, p = 0.033), pelvic rotation (OR = 1.36, p = 0.072), and muscle activation asymmetry (OR = 1.66, p = 0.018). The model demonstrated moderate discriminative ability (AUC = 0.78, 95% CI: 0.64–0.92), sensitivity of 0.75, and specificity of 0.71. However, calibration was suboptimal (slope = 0.24, intercept = 0.47, Brier score = 0.21), suggesting risk underestimation and potential overfitting. Conclusions. Muscle activation asymmetry, cadence, and pelvic kinematic deviations were associated with an increased risk of lower limb injury in high jump athletes. These findings highlight the importance of neuromuscular balance and lumbopelvic stability in injury screening. While the results demonstrate preliminary utility, small sample size and calibration limitations necessitate validation in larger, prospective cohorts before clinical application.