Using Machine Learning Models to Assess Performance, Physiological Indicators, and Injury Risk in Professional Basketball Players
- VernacularTitle:Машин сургалтын загвар ашиглан мэргэжлийн сагсан бөмбөгчдийн гүйцэтгэл, физиологийн суурь үзүүлэлтийг бэртлийн эрсдэлтэй холбон судалсан дүн
- Author:
Jadamba M
1
;
Khaliun M
1
;
Sarangerel D
1
;
Usukhbayar C
1
;
Tsetsegdolgor D
1
;
Burenjargal B
1
;
Odbayasgalan O
2
;
Baasanjargal B
1
;
Ganzorig B
3
;
Otgon G
1
;
Munkhsoyol E
1
Author Information
1. Department of Physiology, School of Biomedical Sciences, MNUMS
2. Mongolian University of Science and Technology
3. Sports Medicine Research Center
- Publication Type:Journal Article
- Keywords:
Conditioning training;
Anaerobic capacity;
Body composition;
Training load
- From:
Mongolian Journal of Health Sciences
2026;91(1):65-71
- CountryMongolia
- Language:Mongolian
-
Abstract:
Background:Basketball involves repeated high-intensity, explosive movements that depend largely on anaerobic energy systems. Accordingly, anaerobic capacity and body-composition parameters such as skeletal muscle mass and fat distribution are key determinants of performance. However, data-driven studies integrating conditioning training, anaerobic performance, and body composition in Mongolian professional players are limited, and the application of machine-learning approaches in this context remains underexplored.
Aim:To examine the relationships between conditioning training, anaerobic capacity, and body composition in professional basketball players and to develop a machine-learning model for injury-risk prediction.
Materials and Methods:Eleven healthy professional male basketball players (21–28 years) from the Omni Erdenet Miners were evaluated. Body composition was measured using the InBody 970, and anaerobic performance was assessed by a 30-s Wingate test on a Monark 894E ergometer. Measurements were obtained at three time points. Changes were analyzed using Friedman and Wilcoxon tests, while associations were examined by Pearson correlation and multivariate regression. Injury-risk models were developed using Random Forest and regression-based machine-learning methods. Research Ethics Committee, approval number 24-25/04-01
Result:Phase angle increased from 6.8±0.5 to 7.3±0.6° (p =0.021), and body-fat percentage decreased from 20.5±3.2% to 16.2±2.0% (p =0.015), with no change in skeletal muscle mass. Visceral fat decreased and anaerobic peak and mean power improved (p <0.05). Fat percentage was negatively correlated with anaerobic performance. Heart-rate recovery improved, indicating enhanced exercise tolerance. Random Forest identified phase angle, visceral fat, fat percentage, and fatigue index as the strongest predictors of injury risk.
Conclusion:1. The body composition of professional basketball players was closely associated with anaerobic performance and recovery indices, with higher body fat percentage and central adiposity tending to be linked to reduced anaerobic capacity.
2. Reducing fat mass while preserving muscle mass did not adversely affect anaerobic performance and demonstrated potential improvements in performance efficiency, load tolerance, and recovery regulation.
3. A Random Forest Regression model based on body composition and anaerobic capacity indicators demonstrated a high level of accuracy in predicting performance decline and injury risk.
- Full text:202609251606552539311_Машин_сургалтын_загвар_ашиглан_мэргэжлийн_сагсан_бөмбөгчдийн.pdf
