Validity of spatiotemporal and ground reaction force estimates during resisted sprinting with a motorized loading device

We aimed to examine the validity of estimating spatiotemporal and ground reaction force (GRF) parameters during resisted sprinting using a robotic loading device (1080 Sprint). Twelve male athletes (age: 20.9 +/- 2.2 years; height: 174.6 +/- 4.2 cm; weight: 69.4 +/- 6.1 kg; means +/- SDs) performed maximal resisted sprinting with three different loads using the device. The step frequency and length and step-averaged velocity, anteroposterior GRF (Fap), and the ratio of Fap to resultant GRF (RF) were estimated using the velocity and towing force data measured using the device. Simultaneously, the corresponding values were measured using a 50-m force plate system. The proportional and fixed biases of the estimated values against those measured using the force plate system were determined using ordinary least product (OLP) regression analysis. Proportional and fixed biases were observed for most variables. However, the proportional bias was small or negligible except for the step frequency. Conversely, the fixed bias was small for step-averaged velocity (0.11 m/s) and step length (0.04 m), whereas it was large for step frequency (0.54 step/s), Fap (16N), and RF (2.22%). For all variables except step frequency, the prediction intervals in the OLP regression dramatically decreased when the corresponding values were smoothed using a two-step moving average. These results indicate that by using the velocity and force data recorded in the loading device, most of the spatiotemporal and GRF variables during resisted sprinting can be estimated with some correction of the fixed bias and data smoothing using the two-step moving average.