Reed Ferber. 1,2. 1. Running Injury Clinic, Calgary, AB. 2. Faculty of Kinesiology, University of Calgary, AB, Canada. Gender Differences in Gait Kinematics in ...
Gender Differences in Gait Kinematics in Competitive and Recreational Runners Angkoon
1,2 Phinyomark ,
1,2 Kobsar ,
1,2 Osis ,
Dylan Sean T. 1,2 1,2 Christian Clermont , Reed Ferber 1Running Injury Clinic, Calgary, AB 2Faculty of Kinesiology, University of Calgary, AB, Canada
Highlights • The purpose of this study was to investigate the differences in running gait kinematics between females and males for recreational and competitive runner subgroups using a principal component analysis (PCA) together with a support vector machine (SVM). • The classification accuracy between genders in a group of competitive runners (86.79%) was significantly higher than a group of recreational runners (79.47%).
• Main differences between males and females in the competitive group were found in sagittal plane knee angles while the differences in the recreational group were found in frontal plane knee angles.
Competitive group
Introduction
Data Collection & Processing
Female runners have a two-fold risk of sustaining certain runningrelated injuries as compared to male runners. It has also been reported that male and female marathon runners are at increased risk of different injuries [1]. Furthermore, the incidence rate for running injuries depends on the specificity of the group of runners concerned (such as recreational joggers versus competitive athletes [2]), which may be due to different extrinsic factors between groups (e.g. running to compete, experience, and training volume [2]). However, few studies have investigated these interactions or differences in running mechanics within these subgroups.
Kinematic joint angles were calculated using four discrete variables of interest during the stance phase [3]: angles at touchdown and toe-off and maximum and minimum peak angles. These gait variables for all three planes of motion, and for three lower extremity joints from both sides, were combined into one 72-dimensional row vector for each subject. A 106-by-72 matrix for a recreational group and a 375-by-72 matrix for a competitive group were created to use as an input for the PCA. The 72 PC scores for each group were sorted based on between-gender group effect size, d, and used as an input for the SVM [3]. A one-way analysis of variance was used to test for statistically significant differences (p