CC BY
6I Journal of Siberian Federal University. Humanities & Social Sciences 2024 17(2): 278-286
EDN: SRIMHW
y^K 004.891.3:796.82
Artificial Intelligence Usage in Prediction ofthe Sports Results of Athletes Competing in Greco-Roman Wrestling
Aleksander Yu. Osipova b c, Roman S. Nagovitsynd e, Tatyana I. Ratmanskaya3, Anna V. Vapaeva3 and Mikhail D. Kudryavtsev*a c f
aSiberian Federal University Krasnoyarsk, Russian Federation
bProf. V.F. Voino-Yasenetsky Krasnoyarsk State Medical University
Krasnoyarsk, Russian Federation
"Siberian Law Institute oftheMinistry oflnternal Affairs
of the Russian Federation Krasnoyarsk, Russian Federation
dGlazov State Pedagogical Institute named after V G. Korolenko
Glazov, Russian Federation
eKazan State Institute of Culture
Kazan, Russian Federation
fSiberian State University of Science and Technology Krasnoyarsk, Russian Federation
Received 24.05.2023, received in revised form 10.10.2023, accepted 08.12.2023
Abstract. Today, the use of machine learning algorithms and neural networks to increase the effectiveness of sports selection at the early stages ofthe athletes' training process is becoming particularly relevant. The aim of this scientific work: to develop a program for predicting the athletic performance of young athletes, who competing in Greco-Roman wrestling, based on artificial intelligence technology.
Collection and processing of individual data of 18-25 years old athletes (n=67) on 21 comparison criteria, ranked into categories in two directions, were implemented: sports space and individual achievements. Two forecasting categories were determined: participants who have obtained a sports title or the highest category (n=16), and participants who have not reached this level (n=17).
The control testing ofthe created program showed only a 14% probability of error in predicting the participants competition performance. According to the functionality of the program in the field of classification of signs by categories, the author's intellectual development with 100 % probability on the basis of experimental approbation revealed key categories of signs that reliably affect the results of the athletes future sports performance.
© Siberian Federal University. All rights reserved
* Corresponding author E-mail address: kumid@yandex.ru, valeo_tei@mail.ru
ORCID: 0000-0002-2277-4467 (Osipov); 0000-0003-4471-0875 (Nagovitsyn); 0000-0001-9544-1674 (Ratmanskaya); 00000002-8081-8974 (Vapaeva); 0000-0002-2432-1699 (Kudryavtsev)
This research defines the main categories of signs that positively or negatively affect the possibility of a wrestler achieving the highest sports rank or title of Russia. Its practical implementation will allow the most accurate, with the minimum level of error in advance to identify athletes predisposed to obtain the highest competition result in Greco-Roman wrestling.
Keywords: combat sports, analysis and forecasting, competition performance, combat athletes, artificial intelligence.
Research area: theory and methodology of sport. Management of athletes' training (planning, modeling, including computer, informational, mathematical, simulation; forecasting, programming, control; scientific and methodological support ofthe training process; discharge standards and qualification requirements).
Citation: Osipov, A. Yu., Nagovitsyn R. S., Ratmanskaya, T.I., VapaevaA. V, Kudryavtsev, M. D. Artificial Intelligence Usage in Prediction of the Sports Results of Athletes Competing in Greco- Roman Wrestling. In: J. Sib. Fed. Univ. Humanit. soc. sci., 2024, 17(2), 278-286. EDN: SRIMHW
Использование искусственного интеллекта
для прогнозирования спортивных результатов атлетов,
соревнующихся в греко-римской борьбе
А.Ю. Осипов3,6,6, P.C. Наговицын Г,А,Т.И. Ратманская\ A.B. Вапаева3, М.Д. Кудрявцев368
"Сибирский федеральный университет Российская Федерация, Красноярск
бКрасноярский государственный медицинский университет имени профессора В. Ф. Войно-ЯсенецкогоМЗ РФ Российская Федерация, Красноярск "Сибирский юридический институтМВДРоссии Российская Федерация, Красноярск
гГлазовский государственный педагогический институт имени В. Г. Короленко
Российская Федерация, Глазов
дКазанский государственный институт культуры
Российская Федерация, Казань
еСибирский государственный университет науки
и технологий имениМ. Ф. Решетнева
Российская Федерация, Красноярск
Аннотация. Сегодня использование алгоритмов машинного обучения и нейронных сетей для повышения эффективности спортивного отбора на ранних этапах тренировочного процесса спортсменов становится особенно актуальным. Цель научной работы: разработать программу прогнозирования спортивных результатов молодых спортсменов, выступающих в греко-римской борьбе, на основе технологии искусственного интеллекта.
Был выполнен сбор и обработка индивидуальных данных спортсменов в возрасте 18-25 лет (п=67) по 21 критерию сравнения, ранжированных по категориям по двум направлениям: спортивное пространство и индивидуальные достижения. Были определены две категории прогнозирования: участники, получившие спортивный титул или высшую категорию (п=16), и участники, которые не достигли этого уровня (п=17).
Контрольное тестирование созданной программы показало лишь 14 %-ную вероятность ошибки в прогнозировании соревновательных результатов участников. Согласно функционалу программы в области классификации признаков по категориям, интеллектуальные разработки автора со 100 % вероятностью на основе экспериментальной апробации выявили ключевые категории признаков, которые достоверно влияют на результаты будущих спортивных выступлений единоборцев. Выводы: в этой научной работе определены основные категории признаков, которые положительно или отрицательно влияют на возможность достижения борцом высшего спортивного разряда или титула в России. Практическая реализация результатов исследования позволит наиболее точно, с минимальным уровнем погрешности, заранее выявить спортсменов, предрасположенных к получению наивысшего соревновательного результата в греко-римской борьбе.
Ключевые слова: спортивные единоборства, анализ и прогнозирование, соревновательный результат, единоборцы, искусственный интеллект.
Научная специальность: 5.8.5 - теория и методика спорта. Управление подготовкой спортсменов (планирование, моделирование, в том числе компьютерное, информационное, математическое, имитационное; прогнозирование, программирование, контроль; научно-методическое обеспечение тренировочного процесса; разрядные нормативы и квалификационные требования).
Цитирование: Осипов А. Ю., Наговицын Р. С., Ратманская Т. И., Вапаева А. В., Кудрявцев М. Д. Использование искусственного интеллекта для прогнозирования спортивных результатов атлетов, соревнующихся в греко-римской борьбе. Журн. Сиб. федер. ун-та. Гуманитарные науки, 2024, 17(2), 278-286. EDN: SRIMHW
Introduction
Recently, artificial intelligence technologies have become an important tool for providing support for the implementation of various processes in the field of physical education (PE) and sports (Liu & Zhu, 2022; Morhat, 2018). When using large data based on intelligent technologies, a systematic analysis of the indicators of PE and sports activities is carried out. Specifically, for optimizing professional and athletic orientation and early identification of athletes with outstanding physical abilities (Bullock et al., 2022; Xu, 2022; Yang & Lin, 2022; Liu et al.; 2021). As well as automation of the quality assessment of technical and tactical preparation and analysis
of information on the results ofthe training process (Nagovitsyn et al., 2023; Qiu et al., 2022; Koliada et al., 2021). Intelligent analytics is beginning to play an important role in improving the quality of PE and athletic training (Tang et al., 2022; Wang & Syed, 2021). At the same time, it is most practically significant through the implementation of informational, methodological and consulting assistance to coaches and PE teachers in various sports organizations (Xu, 2022; Yang & Lin, 2022; Zhang et al., 2022; Wang & Park, 2021).
In specific scientific studies and reports, the use of artificial intelligence (AI) technologies for analyzing large data in the field of sports activities isjustified as one ofthe significant
and effective tools for improving the quality of health and motor indicators of novices and elite athletes (Liu & Zhu, 2022; Li et al., 2021; Morhat, 2018). This makes it possible to solve many problems, such as differentiating those involved to identify dependencies and links between their abilities, analyzing the quality of sports training (Tang et al., 2022; Wang & Syed, 2021) and the risk of achieving low-level physical development (Qiu et al., 2022; Yang & Lin, 2022), automatically building recommendations on the use of information resources and materials to better master the sports training program (Lu, 2022; Minbaleev & Titova, 2020).
In professional sports and fitness, in the last decade, various Al systems with continuous dynamics are beginning to have an increasingly significant impact (Xu, 2022; Koliada et al., 2021; Liu et al., 2021). Intelligent technology is used to analyze large amount of sports data, from the perspective of sports analytics and monitoring referees by video recordings of a match or PA indicators of competitors (Tang et al., 2022; Wang & Syed, 2021; Morhat, 2018). These systems support the physical and athletic performance potential of athletes in the implementation of cyber connections and various virtual sports activities (Dhar, 2017). At various stages of the training and competition process, coaches and athletes use the results of intellectual analytics to personalize training (Yang & Lin, 2022; Wang & Syed, 2021). And also in certain sports situations, Al and machine learning technologies allow you to fully automate the training process (Liu & Zhu, 2022).
However, there are differing opinions on the effectiveness of Al in the fields of PE, sports and fitness (Qiu et al., 2022; Xu, 2022). As some scientific works (Liu et al., 2021; Minbaleev & Titova, 2020) show, the introduction of these technologies without the implementation of experimental studies can be partially negative (Yang & Lin, 2022). As a consequence, it is possible that the next generation of coaches and athletes will be ill-equipped for a dynamic and changing sports world in the space of innovation amid the growing development of large amount of data (Dhar, 2017). However, the approach of this investigation is
not to introduce global standards based on the introduction of intelligent technologies. And also, not in finding ways to replace trainers and coaches in sports in the field of combat sports (Nagovitsyn et al., 2023). With our scientific work, we aim to create such an Al program, in the implementation of which the coach and administration of sports organizations will be able to significantly improve the sports training of combat athletes (Tang et al., 2022; Zhang et al., 2022; Wang & Park, 2021).
Based on the above analysis, a hypothesis was put before the investigation: sports selection for competition performance in wrestling will be more effective if the author's intellectual program for predicting the athletic performance of athletes is created and practically tested. A theoretical analysis of the existing possibilities for using Al to improve the quality of sports training of the younger generation shows that today there is a special need and need for experimental research in this direction. In this regard, the main purpose of this scientific work: to develop a program for predicting the athletic performance of athletes, who competing in Greco-Roman wrestling, based on the Al technology.
Material & methods
At the first stage of this investigation (September-December, 2022) collection and processing of individual data of athletes 18-25 years old (n=67), who competing in Greco-Roman wrestling or already ended wrestling career are carried out. To collect information, an analysis of archival data of athletes from sports schools of three constituent entities of Russian Federation was implemented: the Republic of Tatarstan (Kazan), Krasnoyarsk region (Krasnoyarsk) and the Udmurt Republic (Glazov). Additionally, their personal and telephone survey was carried out. For the investigation, data were collected from athletes during the period when they werejust starting a sports career in Greco-Roman wrestling. The focus of the group included elite athletes, who have the high sports category "Candidate for Master of Sports of Russia" or the sports title "Master of Sports of Russia" (n=16), athletes, who have various sports categories (n=34) and
athletes, who have youth categories or do not have them at all (n=17).
At the second stage of this investigation (January-March, 2023), the development of a forecasting program was implemented. This program is aimed at analyzing the competition performance of participants in the conditions of using Al technologies on the «Orange» platform. Deep neural networks and specially selected machine learning algorithms for categorization were used to carry out the investigation (Nagovitsyn et al., 2023).
For the process of "training the intelligence" of the program on the «Orange» platform, data from combat athletes, who competing in Greco-Roman wrestling, of not the entire experimental sample were used. This process includes data from only athletes selected by special sampling (n=33) for training testing. The main condition for testing the program was the uniformity ofthe samples. So, the tested samples should have included athletes, who having sports ranks and titles (n=16), and not having a sports result at all (n=17).
At the third stage (April-May, 2023), a classification analysis of the developed program was implemented. On the «Orange» intelligent platform, the process of predicting the competition performance of participants was completed and the main classification characteristics for predicting athletic performance were determined. At the final of this process, the program was analyzed using a control sample (n=34) of athletes, who competing in Greco-Roman wrestling and have various sports categories. The control sample data entered into the program made it possible to identify the final percentage of prediction reliability of the author's intellectual development.
In this scientific work, Spearman's rank correlation coefficient and Pearson correlation coefficient were used to determine the degree of dependence of indicators and the relationship between quantitative variables aifecting sports success in Greco-Roman wrestling, where possible.
Based on a comparative analysis of specific scientific knowledge and the peculiarities of sports monitoring, a system of signs and their categories was compiled to implement the
forecast ofthe sports success of Greco-Roman wrestlers (Nagovitsyn et al., 2023; Bullock et al., 2022; Tang et al., Qiu et al., 2022; Xu, 2022; Zhang et al., 2022; Koliada et al., 2021; Liu et al., 2021; Nagovitsyn et al., 2018). This classification was ranked by 21 features, each of which was divided into 2-4 categories. In turn, each in two categories: 12 features by Sports Space and 9 features by Individual Achievements (Table 1).
Further, during the experiment, the individual data of athletes, who competing in Greco-Roman wrestling and participating in the experiment, collected at a time when they were still starting a sports career, were uploaded to the «Orange» analytical system. Based on the processing of this data using deep neural networks and a complex of special machine learning algorithms, two categories of prediction were identified. On the one hand - Greco-Roman wrestlers who have achieved a high sports result (sports title or highest rank). On the other hand, there are Greco-Roman wrestlers who have not reached this level. As a result of analyzing the control sample data, only 2 statistical errors out of 14 tested wrestlers were detected, which corresponds to only 14 % of the error probability of the created author's intellectual program. Thus, the author's intellectual program was tested, finally developed and ready for the implementation of sports selection of athletes, who competing in Greco-Roman wrestling.
Results
By using the functionality of this intelligent program, the categories of signs that most reliably affect the high sports result "Perform KMS/MS" were identified. As a result, the program identified key categories for this result. The identified categories by signs indicate their most reliable significance in the process of implementing the forecast. In turn, the categories of signs that most reliably affect the negative result in Greco-Roman wrestling "Not perform MS/MS" were also revealed. After identifying the main categories of the forecast that most significantly affect the wrestling result (6 categories for each), using the analysis of the data of the entire experimental sample, a mathe-
Table 1. Data classification of Greco-Roman wrestlers by features and their categories
Designation Features Category by coding number
Sports Space
SSI Age ofthe wrestler at the time of arrival in the section 7-8 (0), 9-10 (1), 11-12 (2)
SS2 Marital status ofawrestler Orphan (0), incomplete family (1), complete family (2)
SS3 Wrestler's place ofresidence Village (0), city less than 100 thousand (1), city more than 100 thousand (2)
SS4 The older brother or sister ofthe wrestler Not or not playing sports (0), has a
SS5 Wrestler's father rank (1), has a sports title (2)
SS6 Wrestler's mother
SS7 Coach qualification It has the rank (0), has the rank of KMS or the rank of MS (1), has the rank of MSMK
SS8 Performance ofthe trainer's pupils over the past 5 years Sports rank (0), KMS (1), MS or MSMK rank (2)
SS9 Trainer experience < 5 years (0), 5-15 years (1), 16-30 years (2), > 30 years (3)
SS10 Wrestler's father / wrestler's mother works in sports No (0), yes (1), part-time (2)
SS11 The education of a wrestler School (0), gymnasium (1), lyceum (2)
SS12 Who brought the wrestler to the section Selfor friend (0), grandparents (1), brother or sister (2), father or mother (3)
Individual Achievements
IA1 Average School Score for the Preceding Year < 4,2 (0), 4,2-4,7 (1), > 4,7 (2)
IA2 Skipping training to the number ofall classes (%) > 10 (0), 5-10 (1), < 5 (2)
IA3 Physical development for testing "Ready for work and defense" unsigned (0), bronze/silver (1), gold (2)
IA4 Average performance of a wrestler in the first two or three competitions (place) < 3 (0), 2-3 (1), > 2 (2)
IA5 Cooper's power endurance test (min) < 1,2 (0), 1,2-2 (1),>2 (2)
IA6 Muscle Mass Index to Height and Weight < 15 (0), 15-20 (1), > 20 (2)
IA7 Maximum oxygen consumption < 25 (0), 25-35 (1), > 35 (2)
IA8 Vital capacity oflungs (mL) < 1500 (0), 1500-2000 (1), > 2000 (2)
IA9 Respiratory volume (mL) < 180 (0), 180-250 (1), > 250 (2)
Note: KMS - Candidate for Master of Sports of Russia; MS - Master of Sports of Russia; MSMK - Master of Sports of international class.
matically reliable ratio was obtained in various combinations offeature categories (Table 2).
As a result, the forecast at the level of 100 % is recorded if the wrestler manifests himself during monitoring a combination of 3 categories of signs in one of the different options. In turn, if the athlete's mother goes in for
sports and has a sports title or one of the older brothers or sisters has a sports title, then he is more than 84 % likely to be effective in future competition activities. And if he is additionally coached by a mentor with work experience from 16 to 30 years old or the performance of his previous pupils over the past 5 years will be
Table 2. Categories of features that reliably affect forecasting in various combinations among Greco-Roman style wrestlers
Category offeatures by result "Perform KMS/MS" 1 1-2 1-3
1 Sports mother: holds the sporting title of SS 6(2) / Older brother or sister: has the sporting title of SS 4(2) >84 >91
2 The performance ofthe pupils ofthe coach over the past 5 years: the title ofMS/MSMK SS 8(2) / The experience ofthe coach: from 16 to 30 years SS(2) 100
3 Father or mother works in the field of sports: part-time SS 10(2) / Place ofresidence: city more than 100 thousand SS 3(2)
Category offeatures by result "Not perform KMS/MS" 1 1-2 1-3
1 Who brought the wrestler to the section: grandmother or grandfather SS 1(1) / Wrestling training: Lyceum SS 11(2) >82 >89
2 Average School Score for the Preceding Year: < 4.2 IA1(0) / Omission to Total (%): > 10 IA2(0) 100
3 Physical development on testing "Ready for work and defense": without a sign IA3(0) / Marital status of a wrestler: orphan SS 2(0)
at least the level of MS/MSMK (1 and 2 signs), then the likelihood of reaching the highest sports category and the title of Russia already rises to 91 %.
On the other hand, if a wrestler was "recorded" in the Greco-Roman wrestling section by a grandmother or grandfather or he studied at a moment in an unsportsmanlike lyceum (1 sign), then he later more than 82 % probability will not achieve a high wrestling result in his career. And if he will still additionally miss classes in the Greco-Roman wrestling section or the average score in the school for the previous year will be less than 4.2, then the probability of not reaching the wrestling highest rank and the title of national champion increases to 89 %.
Discussion
In the current period of improving domestic sports training routine, many mathematical methods are used to predict the competition performance of athletes at various stages ofthe training routine (Minbaleev & Titova, 2020; Koptev et al., 2019; Morhat, 2018). In the context of large data analytics, the level of confidence in identifying a significant sports selection strategy begins to increase significantly (Lu, 2022; Wang & Syed, 2021), despite the insufficiency of identifying individual hereditary features and achievements of the athlete (Yang & Lin, 2022; Li et al. 2021; Dhar, 2017).
The implementation of Al programs makes it possible to increase the possibilities of reliable forecasting and significantly increase the accuracy in the processing of athletes' data upon admission to the sports section (Bullock et al., 2022; Liu & Zhu, 2022; Qiu et al., 2022; Xu, 2022). As the present investigation showed, the use of various machine learning algorithms and neural networks made it possible to relatively accurately implement the process of predicting the performance of athletes, who competing in Greco-Roman wrestling, with a minimum level of error. In turn, it is significant to identify athletes, who are not predisposed at the initial stage of the training process to combat sports. The identification of such combat athletes at an early stage of sports training will contribute to further correction ofthe trajectory of educational impact and individualization of the training process with this contingent of athletes (Tang et al., 2022; Nagovitsyn et al., 2019).
Increasing of competition performance in elite sports today is increasingly dependent on the introduction of innovative forms, methods and means and on accurate and timely sports selection even in the early stages of sports activities of novice athletes (Liu et al., 2021; Wang & Syed, 2021). This process is reduced to the search for certain patterns, and as the results of the study showed, combinations of features by categories in comparison with
the indicators of the entire study group (Nagovitsyn et al., 2023). Which, in turn, can be reduced to the task of systematization by the signs of comparing the indicators of combat athletes and the final sports results as predictive models based on large amount data analysis (Tang et al., 2022; Xu, 2022). The created program on intellectual patterns according to the characteristics of comparison reached a high level of 100 % with a combination of minimum three characteristics. And if only one category of the sign coincides, the probability was 84 % to achieve the wrestling category Candidate for Master of Sports of Russia or the title Master of Sports of Russia and 82 % to not fulfill this sports level.
Conclusions
Thus, in scientific work, an author's program based on artificial intelligence was created and tested to implement the forecast of the competitive performance of Greco-Roman martial arts. The study statistically defines the main categories of signs that positively or negatively affect the possibility of a wrestler achieving the highest sports rank or title of Russia. Its practical implementation will allow the most accurate, with the minimum level of error in advance to identify athletes predisposed to obtain the highest result in Greco-Roman wrestling. As the statistical results of the study showed, the introduction ofthe functionality of neural networks and a specially selected complex of machine learning algorithms increases the quality of sports selection in Greco-Roman wrestling. This ultimately allows you to timely individualize and improve the training process
References
of wrestlers. In this view, it is intelligent technologies using special algorithms for the implementation of forecasting that have effective and reliable practical significance in the process of sports selection. Especially to the ability of mining, the processes of classification and identification of significant data that are statistically reliable to increase the efficiency of the sports trek of athletes, who competing in Greco-Roman wrestling.
Limitations
Any inferences made in this scientific work have some important limitations. These limitations associated with the total low number of participants (non-elite and elite athletes, who competing in Greco-Roman wrestling). A possible limitation is also the lack of data of athletes, who competing in Freestyle wrestling, which does not allow comparing, evaluating and analyzing the factors of sporting success in Freestyle and Greco-Roman wrestling.
Acknowledgements
We would like to thank the personnel of Sports wrestling academy named D. G. Mindi-ashvili (Krasnoyarsk) and Sport Greco-Roman schools of Olympic reserve (Glazov, Kazan and Krasnoyarsk), all athletes, their coaches and sports professionals, who participated in this scientific work.
Conflicts of interest
The authors declare that this scientific work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Bullock G. S., Hughes T., Arundale A. H., Ward P., Collins G. S., Kluzek S. Black box prediction methods in sports medicine deserve a red card for reckless practice: A change of tactics is needed to advance athlete care. In Sports Medicine, 2022, 52(8), 1729-1735. doi:10.1007/s40279-022-01655-6
Dhar V. What is the role of artificial intelligence in sports? \nBig Data, 2017,5(3), 173-174. doi: 10.1089/ big.2017.29022.vdb
Koliada M. G., Belykh S. I., Bugayova T. I., Oleinik O. S. Artificial intelligence method to detect psychological and pedagogical anomalies in physical education and sports activities. In Teoriya i Praktika Fizicheskoy Kultury, 2021, 9, 66-69.
Koptev O. V., Osipov A. Y., Kudryavtsev M. D., Zhavner T. V., Klimuk Y. V., Vapaeva A. V., Kuz-min V. A., Mokrova T. I. Level increase of competitive readiness of elite judokas in the weight category of
up to 60 kg (as an example is the national team of Kyrgyzstan). YnJournal of Physical Education and Sport, 2019, 19(1), 716-721. doi:10.7752/jpes.2019.01103
Li L., Zhang L., Zhang S. Using artificial intelligence for the construction of university physical training and teaching systems. In Journal Healthcare Engineering. 2021, 29, 3479208. doi: 10.1155/2021/3479208 Liu H., Zhu X. Design of the physical fitness evaluation information management system of sports athletes based on artificial intelligence. In Computational Intelligence and Neuroscience, 2022, 30, 1925757. doi:10.1155/2022/1925757
Liu T., Wilczyñska D., Lipowski M., Zhao Z. Optimization of a sports activity development model using artificial intelligence under new curriculum reform. In International Journal of Environmental Research and Public Health, 2021, 18(17), 9049. doi:10.3390/ijerphl8179049
Lu G. Prediction model and data simulation of sports performance based on the artificial intelligence algorithm. In Computational Intelligence and Neuroscience, 2022, 7, 7238789. doi: 10.1155/2022/7238789
Minbaleev A. V., Titova E. V. The use of artificial intelligence in sports and its legal limitations. In Human. Sport. Medicine, 2020, 20(S 2), 114-119. doi:10.14529/hsm20s218
Morhat P.M. Artificial intelligence in the field of sports: opportunities, directions and methods of involvement. InTeoriyalpraktikafizicheskoyKultury, 2018, 10, 95-97.
Nagovitsyn R. S., Gibadullin I. G., Batsina O.N., Mokrushina I. A. Forecasting the competitive performance of young athletes based on artificial intelligence technology. In Teoriya I praktika fizicheskoy Kultury, 2023, 2, 27-30.
Nagovitsyn R. S., Osipov A. Y., Kapustin A. G., Anfilatova O. V., Senator S. Y. Determination of the dependence of competitive results on the procedure of sports selection among Greco-Roman wrestling athletes. In Pedagogy of Physical Culture and Sports, 2019, 23(4), 182-188. dot: 10.15561/18189172.2019.0404 Nagovitsyn R. S., Zhuikova S. E., Kondratiev N. V., Osipov A. Y., Zhavner T. V., Vapaeva A. V. Influence of sports asymmetry and ambidexterity of ground wrestling on the level of competitive performance of Greco-Roman style wrestlers. In Journal of Physical Education and Sport, 2018, 18(4), 2472-2477. doi:10.7752/jpes.2018.04370
Qiu C., Su C., Liu X., Yu D. A study of feature construction based on least squares and RBF neural networks in sports training behaviour prediction. In Computational Intelligence and Neuroscience, 2022, 7, 5034081. dot: 10.1155/2022/5034081
Tang H., Jiang G., Wang Q. Prediction of college students' sports performance based on improved BP neural network. In Computational Intelligence and Neuroscience, 2022, 8, 5872384. doi: 10.1155/2022/5872384 Xu J. Prediction and planning of sports competition based on deep neural network. In Computational Intelligence and Neuroscience, 2022, 8, 1906580. doi:10.1155/2022/1906580
Wang J., Syed H. A. Analysis of sports performance prediction model based on GA-BP neural network algorithm. In Computational Intelligence and Neuroscience, 2021, 12, 4091821. doi: 10.1155/2021/4091821
Wang T., Park J. Design and implementation of intelligent sports training system for college students' mental healtheducation. In Frontiers in Psychology, 2021, 12,634978. doi:10.3389/fpsyg.2021.634978
Yang R., Lin H. Evaluation of the quality of football teaching in colleges and universities based on artificial neural networks. In Computational Intelligence and Neuroscience, 2022, 27, 8001252. doi: 10.1155/2022/8001252
Zhang B., Jin H., Duan X. Physical education movement and comprehensive health quality intervention under the background of artificial intelligence. In Frontiers in Public Health, 2022, 10, 947731. doi:10.3389/fpubh.2022.947731
