(71) ESTABLISHING THE NEUROMUSCULAR BENCHMARK: AN UNSUPERVISED MACHINE LEARNING ANALYSIS EXAMINING MUSCLE STRENGTH IN FEMALE ATHLETES

Purpose: There are currently no universally accepted approaches for determining neuromuscular strength benchmarks in female university athletes to evaluate performance, identify athletes at increased risk for sport injury, and inform return to sport readiness. The objectives of this study are to determine if distinct neuromuscular strength profiles exist in female university athletes; and examine the influence of biological and sociological (bio-social) factors on neuromuscular strength profiles. Methods: Female university athletes (n=192) from five sports completed a lower body neuromuscular strength testing battery to quantify maximal strength, power, and plyometric capacities, as well as a questionnaire assessing attitudes towards strength training and injury prevention. An unsupervised machine learning clustering analysis was completed to cluster neuromuscular strength data and Fisher’s Exact Tests were used to examine differences in attitudes toward training across clusters. Results: Three clusters representing unique neuromuscular performance profiles were identified. One cluster showed greater maximal strength and faster jump contraction times compared to the other two. Subtle differences in athlete attitudes towards training between clusters was noted (p< 0.047) with the strong cluster being the only group to not include athletes with extreme negative attitudes towards the benefits of strength training. Conclusions: An unsupervised machine learning analysis identified three neuromuscular profiles in female university athletes. Although differences in attitudes towards strength training between clusters were observed, the potential relationship between biological and social factors needs further exploration. Practical Applications: Establishing a data-driven approach to neuromuscular benchmarking for this population is critical not only for evaluating athlete performance and to inform injury rehabilitation. For instance, ACL RTS guidelines commonly rely on the uninvolved contralateral limb as a benchmark of neuromuscular function which is limited, and a data driven benchmark provides opportunities to examine global athlete performance profiles that may improve return to sport outcomes. Acknowledgements: None