Predicting high bar forces in the longswing

The longswing on high bar in men's artistic gymnastics is the core skill within all competitive routines. The forces applied to the bar by the gymnast are important when studying a performer's technique or when examining injury mechanisms. Previous studies have used video measurements of the bar's motion to predict bar forces to within 7% of the range of directly recorded forces. Also, by assuming zero external forces at the gymnast's feet, inverse dynamics have been used to predict bar forces, but previously this method has resulted in errors greater than 20%. A study, employing 2D DLT techniques and customized inertia data for four elite male gymnasts performing three longswings on a strain gauged high bar, was conducted to enable the two methods for estimating bar forces to be directly compared. Digital video images were recorded at 50Hz from which the bar centre, head centre and the nearest wrist, elbow, shoulder, hip, knee, ankle and toe were digitized, starting before the gymnast reached the handstand, continuing throughout one revolution and ending once the gymnast passed beyond the handstand (∼400°). All video and force data (l000Hz) were interpolated within a single 360° longswing at 1° intervals and root mean squared differences (rmsd) between the measured bar forces and those predicted by bar deformation and inverse dynamics were compared. Tracking the motion of the bar in 2D was poor in comparison to the 8% rmsd when using inverse dynamics. In the latter technique, deliberately swapping inertia data sets between the subjects increased errors. Inverse dynamics data were sensitive to kinematic and inertia data errors but the use of the 2D DLT and the inclusion of personalized body segment parameters contributed to an overall reduction in error compared to previously reported data. When direct bar force measurement cannot be obtained, the bar deformation technique is recommended providing that 3D video is used with a pre calibrated bar. Alternatively, with appropriate inertia data and DLT processing, the inverse dynamics technique can be employed, albeit with a slight loss of overall accuracy in predicting the precise profile of the high bar forces.