Related Experiment Video
Updated: Aug 24, 2025
06:52
An Inertial Measurement Unit Based Method to Estimate Hip and Knee Joint Kinematics in Team Sport Athletes on the Field
Published on: May 26, 2020
8.0K
Tracking Health, Performance and Recovery in Athletes Using Machine Learning
Denis V Petrovsky1, Vasiliy I Pustovoyt2, Kirill S Nikolsky1
1Biobanking Group, Branch of Institute of Biomedical Chemistry "Scientific and Education Center", 109028 Moscow, Russia.
Sports (Basel, Switzerland)
|October 26, 2022
Summary
Athlete recovery post-competition can be assessed by analyzing blood and urine biochemistry. Key indicators of muscle metabolism and the ornithine cycle effectively differentiate catabolism and anabolism phenotypes.
Area of Science:
- Sports Medicine
- Biochemistry
- Data Science
Background:
- Intense athletic training and competition can lead to temporary performance impairments, with recovery times varying from days to longer periods.
- Comprehensive health assessments are crucial for understanding athlete well-being and recovery.
- Identifying biomarkers for metabolic states is essential for optimizing athlete recovery strategies.
Purpose of the Study:
- To identify significant biochemical indicators in athlete blood and urine that characterize catabolism and anabolism phenotypes.
- To evaluate the effectiveness of machine learning models in analyzing these indicators for recovery assessment.
- To understand the metabolic recovery processes in athletes during the post-competitive period.
Main Methods:
- Analysis of health indicators from 3661 athletes undergoing in-depth medical examinations.
- Utilized instrumental (fluorography, ultrasound, echocardiography, ECG, stress testing) and laboratory (urinalysis, blood tests) examinations.
- Applied machine learning methods, specifically random forest and multinomial logistic regression, to identify key biochemical markers.
Main Results:
- Machine learning models successfully identified significant blood and urine biochemistry indicators for classifying catabolism and anabolism phenotypes.
- Muscle metabolism parameters (aspartate aminotransferase, creatine kinase, lactate dehydrogenase, alanine aminotransferase) were highly significant.
- Ornithine cycle parameters (creatinine, urea acid, urea) also significantly contributed to metabolic phenotype classification.
Conclusions:
- Biochemical markers of muscle metabolism and the ornithine cycle are crucial for characterizing athlete metabolic states post-competition.
- Machine learning approaches provide effective tools for analyzing these indicators to assess recovery effectiveness.
- This study offers insights into optimizing recovery strategies by understanding individual athlete metabolic responses.
