Predicting Running-Related Injuries from Functional, Kinetic and Kinematic Data.

Abstract

The literature has identified inconsistent biomechanical risk factors for running-related injuries but lacks investigations on interactions between biomechanics and other risk factors. This prospective cohort study aimed to develop and compare prediction models of various levels of complexity to predict running-related injuries over 12 months in recreational runners. The seven-item functional movement screen test was administered at baseline for 83 participants. Running biomechanics were evaluated using clinically friendly tools, including wearable in-shoe force sensors to measure vertical ground reaction forces and 2D video-based kinematic analysis of lower extremities. The participants were subsequently monitored over a 12-month follow-up period to track whether they sustained running-related injuries. Differences between the injured (n=26) and non-injured (n=55) groups were examined using the Mann-Whitney U-test. Binary logistic regression was performed to identify significant indicators for running-related injuries, with six models developed involving different sets of variables. Neither simple (involving one variable) nor complex models (including multiple variables) were statistically significant (p-values ranged from 0.106 to 0.972). In conclusion, prediction models developed using variables obtained from accessible tools are unable to accurately predict future running-related injuries regardless of model complexity. Researchers and practitioners should avoid over-reliance on simple measures for screening injury risks.