MACHINE LEARNING AS A PROMISING TOOL FOR BRIDGING THE GAP IN WOMEN TEAM SPORTS INJURY PREVENTION

In recent years, the participation, professionalism, and success of women in sports has increased exponentially. However, there is a dangerous gap in the development and popularization of women's sports due to the lack of sports science and sports medicine research on elite, sub-elite and amateur female athletes. Moreover, applying findings from sports science obtained on male athletes to female athletes can be flawed. Not only inappropriate training loads based on male sports science, but also disregard for physiological and biological differences can have a negative impact and lead to injury. Current mono-dimensional approach for injury prevention and prediction, based on screening tests and protentional programs, is not effective in practice due to low precision. Therefore, female team sports urgently need an alternative approach based on Machine Learning (ML), an easy-to-use tool, that is able to detect risk factors at an early stage, in order to decrease overall injury incidence and cost. The overall objective of this project is to develop a ML tool for female team sports that is able to: 1) predict the occurrence of an injury using pre-season fitness, neuromuscular and stress parameters; 2) identify which parameter(s) contribute the most and represent an injury risk factor(s) at an individual level. This project will be based on a mass measurement cross-sectional study design in order to obtain the athletes’ fitness, neuromuscular and stress parameters, as well as a prospective cohort design related to the epidemiology of sports injuries. The project will be focused on female team sports and include junior, adolescent, and senior female athletes competing at different levels (professional, semi-professional, amateur). We will initially target at least five teams for each sport (football, handball, basketball, volleyball); therefore 20 team sport clubs with around 20-250 female participants in total will be included. ML algorithms suitable for supervised learning tasks (i.e., decision trees, regression models, random forest and XGBoost) will be used for modelling. The developed ML tool for injury prevention and prediction could enable us to connect, interact and share knowledge with the Medical Networks (FIFA, FIBA, FIVB etc. medical centers) by offering a new and effective tool. We will dedicate this project exclusively to female team sports athletes in order to diminish the gap in the scientific literature and make scientific evidence applicative in the practice to maximize performance potential of women cohort.