Can Running Distance Variables in football Predict the occurrence of sport injuries of the Iranian Premier League players?
Poster Presentation
Paper ID : 1165-SSRC
Authors
1Health and sports rehabilitation, Faculty Sport Sciences and Health, Shahid Beheshti University, Tehran, Iran
2Assistant Professor Department of Pathology and Rehabilitation, Faculty of Physical Education and Sport Sciences, Shahid Beheshti University
3Assistant Professor, Department of Pathology and Rehabilitation, Faculty of Physical Education and Sport Sciences, Shahid Beheshti University
4HEME Research Group, Faculty of Sport Sciences, University of Extremadura
Abstract
Background: The management of training load stands as a critical external risk factor influencing football injuries. Within the array of exercise load indicators, distance covered, an aspect of external training load, remains relatively underexplored within injury prediction models. This study aimed to examine the specific role of distance covered in predicting sports injuries among football players in Iran's premier league, utilizing the decision tree algorithm.
Methodology: Twenty-five players from a team in Iran's Premier Football League during the 2014-2015 season, averaging 26.1±3.2 years in age, 77.5±7.6 kg in weight, 1.82±0.07 m in height, and with a body mass index of 23.4±1, participated in this research. GPS data and player injuries throughout an entire season were prospectively recorded by analytical and medical personnel. Subsequently, the acute-to-chronic workload ratio for GPS data was computed using the Uncapold method over a 7 to 21-day span. then, the machine learning algorithm employed a decision tree methodology utilizing two Indicators: the ratio of acute to chronic workload for each parameter in the injured player, and the average ratio of acute to chronic workload across all players in the week preceding the injury event. This approach aimed to assess the predictive potential of these indices in determining recorded injuries.
Results: Implementation of the decision tree algorithm, both in Indicator aggregation and Indicator averaging, revealed four indicators. In running the decision tree algorithm on the test data, the total distance covered demonstrated an AUC of 0.73, Recall of 50%, Precision of 50%, and Accuracy of 92.9%. Meanwhile, the average distance covered demonstrated an AUC of 0.74, Recall of 50%, Precision of 57.1%, and Accuracy of 93.8%.
Conclusion: The research findings highlighted the moderate predictive efficacy of both Indicators. Therefore, it is recommended for football coaches and managers to consider the acute distance Indicator covered by a player in a single session and the acute-to-chronic ratio of this Indicator as pivotal indicators in predicting injuries among football players.
Methodology: Twenty-five players from a team in Iran's Premier Football League during the 2014-2015 season, averaging 26.1±3.2 years in age, 77.5±7.6 kg in weight, 1.82±0.07 m in height, and with a body mass index of 23.4±1, participated in this research. GPS data and player injuries throughout an entire season were prospectively recorded by analytical and medical personnel. Subsequently, the acute-to-chronic workload ratio for GPS data was computed using the Uncapold method over a 7 to 21-day span. then, the machine learning algorithm employed a decision tree methodology utilizing two Indicators: the ratio of acute to chronic workload for each parameter in the injured player, and the average ratio of acute to chronic workload across all players in the week preceding the injury event. This approach aimed to assess the predictive potential of these indices in determining recorded injuries.
Results: Implementation of the decision tree algorithm, both in Indicator aggregation and Indicator averaging, revealed four indicators. In running the decision tree algorithm on the test data, the total distance covered demonstrated an AUC of 0.73, Recall of 50%, Precision of 50%, and Accuracy of 92.9%. Meanwhile, the average distance covered demonstrated an AUC of 0.74, Recall of 50%, Precision of 57.1%, and Accuracy of 93.8%.
Conclusion: The research findings highlighted the moderate predictive efficacy of both Indicators. Therefore, it is recommended for football coaches and managers to consider the acute distance Indicator covered by a player in a single session and the acute-to-chronic ratio of this Indicator as pivotal indicators in predicting injuries among football players.
Keywords