Analyzing Force-Time Curves: Comparison of Commercially Available Automated Software and Custom MATLAB Analyses

Abstract

Merrigan, JJ, Stone, JD, Galster, SM, and Hagen, JA. Analyzing force-time curves: Comparison of commercially available automated software and custom MATLAB analyses. J Strength Cond Res 36(9): 2387-2402, 2022-With the growing prevalence of commercial force plate solutions providing automated force-time curve analysis, it is critical to understand the level of agreement across techniques. Thus, this study directly compared commercial and custom software analyses across force-time curves. Twenty-four male and female subjects completed 6 trials of countermovement, squat, and drop jumps, and isometric mid-thigh pulls on the same force plate. Vertical ground reaction forces were analyzed by automated software from Vald Performance, Hawkin Dynamics, and custom MATLAB scripts. Trials were visually assessed to verify proper landmark identifications. Systematic and proportional bias among analyses were compared via least products regressions, Bland-Altman plots, and percent error. Hawkin Dynamics had subtle differences in analysis procedures and demonstrated low percent errors across all tests (<3% error), despite demonstrating systematic and proportional bias for several metrics. ForceDecks demonstrated larger percent differences and greater biases for several metrics. These errors likely result from different identification of movement initiation, system weight, and integration techniques, which causes error to subsequent landmark identifications (e.g., braking/propulsive phases) and respective force-time metrics. Many metrics were in agreement between devices, such as isometric mid-thigh pull peak force consistently within 1 N across analyses, but some metrics are difficult and incomparable across software analyses (i.e., rate of force development). Overall, many metrics were in agreement across each commercial software and custom MATLAB analyses after visually confirming landmarks. However, because of inconsistencies, it is important to only compare metrics that are in agreement across software analyses when absolutely necessary.