Determining the Level of Reaction to a Moving Object in Combat Sports Athletes Using a Composite Index and Machine Learning: A Cross-Sectional Methodological Study.
6 affiliations
- 1Department of Martial Arts, Kharkiv State Academy of Physical Culture, 61058 Kharkiv, Ukraine.
- 2Department of Theory and Methodology of Physical Education, Uzbek State University of Physical Education and Sport, Chirchiq 111700, Uzbekistan.
- 3Department Combat Sport, National University of Ukraine on Physical Education and Sport, 03150 Kyiv, Ukraine.
- 4Department of Sports Psychology and Pedagogy, Uzbek State University of Physical Education and Sport, Chirchiq 111700, Uzbekistan.
- 5Institute of Physical Education and Sport, Technical University in Zvolen, 96001 Zvolen, Slovakia.
- 6Department of Olympic and Professional Sports, Kherson State University, 76018 Ivano-Frankivsk, Ukraine.
Abstract
Background. Reaction to a moving object (RMO) is an important psychophysiological function in combat sports, supporting situational perception, anticipation of an opponent's movements, and timely technical and tactical actions. Objectives. This study aimed to assess this function in combat sports athletes using a composite index and to develop a machine learning model for automated classification. Methods. The study included 312 combat sports athletes aged 10-25 years, approximately 10% of whom were female, representing five combat-sport disciplines and a broad range of sports qualifications from beginners to highly qualified athletes. Reaction to a moving object was assessed using the specialized "Reaction RMO Pro" test. The RMO_Index was calculated from standardized measures of mean absolute temporal deviation and its variability, with higher values indicating better overall performance. The model was trained using stage-specific test indicators and characteristics of premature, delayed, and accurate responses. Results. Mean absolute temporal deviation was 32.44 ± 10.29 ms, and variability was 21.88 ± 6.87 ms. The proportions of premature, delayed, and accurate responses were 48.18 ± 10.18%, 44.66 ± 10.45%, and 7.16 ± 4.40%, respectively. A balanced target variable, RMO_Level_Target, was derived from the RMO_Index: High, 104; Medium, 104; and Low, 104. Within the training-validation dataset, Create ML reported 100% classification accuracy during model development, while the model achieved 91.67% accuracy on the held-out test subset. The F1-scores were 0.93, 0.94, and 0.88 for the High, Low, and Medium classes, respectively. Conclusions. The RMO_Index provided a standardized composite measure of response timing accuracy and stability, and enabled the formation of sample-derived High, Medium, and Low performance levels. The Boosted Tree Classifier reproduced these reference levels with 91.67% accuracy in the held-out test subset. These findings support the feasibility of automated classification of RMO test results within the present sample; however, external validation is required before broader practical application.
Related
- Short-term Rhodiola rosea supplementation improves premenstrual symptoms, reaction time, and perceived exertion in female taekwondo athletes: a randomised double-blind placebo-controlled trial
- Determining the cognitive function differences between different types of physically active students : Cross-sectional analysis
- Effects of Yeast Protein–Whey Protein Composite Supplementation on Recovery from Exercise-Induced Muscle Damage in Well-Trained Athletes: A Randomized Controlled Trial
- Statistical Parametric Mapping reveals sex-specific, phase-dependent differences in bilateral three-dimensional ground reaction force during Wushu Changbing PiJi in high-level athletes
- Psychometric properties and cross-cultural adaptation of the HPV testing knowledge scale: a cross-sectional methodological study among Turkish women
BibTeX
@article{romanenko2026determining,
title = {{Determining the Level of Reaction to a Moving Object in Combat Sports Athletes Using a Composite Index and Machine Learning: A Cross-Sectional Methodological Study.}},
author = {Vyacheslav Romanenko and Yrui Tropin and Rashid Matkarimov and Georgiy Korobeynikov and Fikrat Kerimov and Lesia Korobeinikova and Shukurjon Gaziev and Olha Zadorozhna},
journal = {Journal of functional morphology and kinesiology},
year = {2026},
month = oct,
doi = {10.3390/jfmk11040431},
url = {https://europepmc.org/article/MED/42924710},
note = {CC BY},
}
RIS
TY - JOUR
TI - Determining the Level of Reaction to a Moving Object in Combat Sports Athletes Using a Composite Index and Machine Learning: A Cross-Sectional Methodological Study.
AU - Vyacheslav Romanenko
AU - Yrui Tropin
AU - Rashid Matkarimov
AU - Georgiy Korobeynikov
AU - Fikrat Kerimov
AU - Lesia Korobeinikova
AU - Shukurjon Gaziev
AU - Olha Zadorozhna
PY - 2026
DA - 2026/10/18
JO - Journal of functional morphology and kinesiology
DO - 10.3390/jfmk11040431
UR - https://europepmc.org/article/MED/42924710
AB - Background. Reaction to a moving object (RMO) is an important psychophysiological function in combat sports, supporting situational perception, anticipation of an opponent's movements, and timely technical and tactical actions. Objectives. This study aimed to assess this function in combat sports athletes using a composite index and to develop a machine learning model for automated classification. Methods. The study included 312 combat sports athletes aged 10-25 years, approximately 10% of whom were female, representing five combat-sport disciplines and a broad range of sports qualifications from beginners to highly qualified athletes. Reaction to a moving object was assessed using the specialized "Reaction RMO Pro" test. The RMO_Index was calculated from standardized measures of mean absolute temporal deviation and its variability, with higher values indicating better overall performance. The model was trained using stage-specific test indicators and characteristics of premature, delayed, and accurate responses. Results. Mean absolute temporal deviation was 32.44 ± 10.29 ms, and variability was 21.88 ± 6.87 ms. The proportions of premature, delayed, and accurate responses were 48.18 ± 10.18%, 44.66 ± 10.45%, and 7.16 ± 4.40%, respectively. A balanced target variable, RMO_Level_Target, was derived from the RMO_Index: High, 104; Medium, 104; and Low, 104. Within the training-validation dataset, Create ML reported 100% classification accuracy during model development, while the model achieved 91.67% accuracy on the held-out test subset. The F1-scores were 0.93, 0.94, and 0.88 for the High, Low, and Medium classes, respectively. Conclusions. The RMO_Index provided a standardized composite measure of response timing accuracy and stability, and enabled the formation of sample-derived High, Medium, and Low performance levels. The Boosted Tree Classifier reproduced these reference levels with 91.67% accuracy in the held-out test subset. These findings support the feasibility of automated classification of RMO test results within the present sample; however, external validation is required before broader practical application.
SN - 2411-5142
ER -