Algorithmic Audits in Sports Medicine: An Examination of the SpartaScience™ Force Plate System.

INTRODUCTION

Force plate systems are increasingly utilized in the armed forces that claim to identify individuals at risk of musculoskeletal injury. However, factors influencing injury risk scores from a force plate system (SpartaScience™) and the effects of experimental perturbations on these scores remain unclear.

METHODS

Healthy males ( n = 823; 22.7 ± 3.9 yr) performed a countermovement jump (CMJ) on SpartaScience™ force plates. Multiple regression analysis was used to identify predictors of the system's proprietary Musculoskeletal (MSK) Health score, which were then experimentally perturbed. Twelve males (30.9 ± 4.3 yr) participated in a test-retest reliability study, performing three standard CMJs and one experimentally manipulated jump (50% effort) due to the observed relationship between the MSK Health score, vertical jump height, and body weight.

RESULTS

The MSK Health score was negatively correlated with vertical jump height and positively with body weight ( R2 = 0.59, P < 0.001). Each inch increase in jump height decreased the MSK Health score by an average of 1.27 units (95% confidence interval, 1.17-1.36), whereas each pound of body weight increased it by 0.12 units (95% confidence interval, 0.11-0.13). Notably, 83% of participants in the reliability study improved their MSK Health score on the 50% effort jump.

CONCLUSIONS

The study revealed atypical relationships between MSK Health scores, vertical jump height, and body weight, with vertical jump height playing a majority role in predicting the principal output (MSK Health score). Findings indicated a higher injury risk with greater jump height but a lower risk with increased body weight. In addition, MSK Health scores paradoxically improved with lower effort (i.e., lower vertical jump height), which highlights the dangers of using undisclosed and unvetted algorithms for the prediction of health outcomes.