(40) Novel Machine Learning Approach to Predict Jump Performance from Interlimb Asymmetry in Athletes

Purpose: Understanding an athlete’s fitness-fatigue status is crucial for optimizing their sport-specific performance capabilities. So, access to data that identifies neuromuscular fatigue may help sports practitioners to make data-driven and effective decisions in managing an athlete’s workloads. The purpose of this study was to identify variables about human movement at will allow for early detection of neuromuscular fatigue. Methods: Eleven females (X±SD; age=20.8±1.1 yrs; hgt=72.2±7.4 cm; wgt=68.0±7.2 kg) and eleven males (age=23.0±2.6 yrs; hgt=180.3±4.8 cm; wgt=80.4±7.3 kg) volunteered to participate in the present investigation. Participants were screened using a standardized movement screening protocol, consisting of 5 lower body jumping tasks. A three-dimensional markerless motion capture system (DARI Motion Inc, Overland Park, KS) was used to analyze the kinetic and kinematic variables of interest that were calculated and reported. Each participant completed one familiarization session (baseline) and an acute fatiguing protocol [i.e. BOSCO; repeated vertical jump test (RVJT)]. An ANOVA was used for statistical analysis comparison. Results: Data showed that the pre and post-test analysis indicated the reactive strength index (RSI) (baseline = 2.7, +/- 0.5 ; post = 1.7, +/- 0.4), contact flight time ratio during multi hop (baseline = 1.9, +/- 0.6 ; post = 0.9, +/- 0.3), and unilateral jump net impulse (baseline = 248Ns +/- 37Ns ; post = 192Ns +/- 44Ns) were statistically significant from baseline. Conclusions: The identified variables are affected by neurological fatigue while other variables were not suggesting that they could be used for more sensitive monitoring of athlete fatigue. Practical Applications: A narrower and more simplified test can be designed and administered to help understand if an athlete is neuromuscular fatigued and alternation to their training came be made more rapidly. Acknowledgements: None