Scholarlink

The lack of training to maximize the patient’s physical performance, especially in weightlifting, can cause various types of musculoskeletal injuries. The biomechanical analysis through motion sensors and machine learning algorithms of an exercise as widely used in any sports discipline as the free squat can be a very helpful tool for both beginners and high-performance athletes. This methodology, using Xsens DOT© motion sensors which internally contain a gyroscope, records the exercise. With the necessary knowledge, the data is processed using feature extraction with statistical moments, allowing for the classification of the exercise execution with prior knowledge in sports and basic anatomy. Additionally, genetic algorithms are used to select the best variables to create a logistic classification model, achieving good performance. The proposed model obtained an accuracy of 0.7885 in a blind evaluation and an area under the receiver operating characteristic curve (AUC) of 0.74.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Analysis of the Correct Execution of Free Squats for Injury Prevention: A Genetic Algorithm Approach

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Sebastian Guzmán-Alfaro,
  • José I. Ríos-Ríos,
  • José M. Celaya-Padilla,
  • Carlos E. Galván-Tejada,
  • Angeles O. Félix-Castañeda,
  • Manuel A. Soto-Murillo,
  • Eduardo de Avila-Armenta,
  • Pedro E. Mayorga-Ruiz,
  • Andrea Acuña-Correa

摘要

The lack of training to maximize the patient’s physical performance, especially in weightlifting, can cause various types of musculoskeletal injuries. The biomechanical analysis through motion sensors and machine learning algorithms of an exercise as widely used in any sports discipline as the free squat can be a very helpful tool for both beginners and high-performance athletes. This methodology, using Xsens DOT© motion sensors which internally contain a gyroscope, records the exercise. With the necessary knowledge, the data is processed using feature extraction with statistical moments, allowing for the classification of the exercise execution with prior knowledge in sports and basic anatomy. Additionally, genetic algorithms are used to select the best variables to create a logistic classification model, achieving good performance. The proposed model obtained an accuracy of 0.7885 in a blind evaluation and an area under the receiver operating characteristic curve (AUC) of 0.74.