Mehmet Fatih AKAY, Fatih ABUT, İmdat YARIM, Boubacar SOW, Ebru ÇETİN

Support vector machines for predicting the hamstring and quadriceps muscle strength of college-aged athletes

Hamstring and quadriceps muscles are essential for the performance of athletes in various sport branches. Hamstring muscles control running activities and stabilize the knee during turns or tackles, while quadriceps muscles play an important role in jumping and kicking. Although hamstring and quadriceps muscle strength in athletes can be accurately measured using isokinetic dynamometry, practical difficulties, such as the requirement of nonportable and costly equipment as well as a long period of measurement time, motivate the researcher to predict hamstring and quadriceps muscle strength using promising machine-learning methods. The purpose of this study is to build prediction models for estimating the hamstring and quadriceps muscle strength of college-aged athletes using a support vector machine (SVM). The data set included 75 athletes selected from the College of Physical Education and Sport, Gazi University, Turkey. The predictor variables of sex, age, height, weight, body mass index, and sport branch were utilized to build the hamstring and quadriceps muscle strength prediction models for various types of training methods. The generalization error of the prediction models was calculated by carrying out 10-fold cross-validation, and the prediction errors were evaluated using several performance metrics. For comparison purposes, prediction models based on a radial basis function neural network (RBFNN) and single decision tree (SDT) were also developed. The results reveal that the SVM-based hamstring and quadriceps strength prediction models significantly outperform the RBFNN-based and SDT-based models and can be safely utilized to produce predictions regarding new data with acceptable accuracy.

PDF

___

[1] Di Michele R, Del Curto L, Merni F. Mechanical and metabolic responses during a high-intensity circuit training workout in competitive runners. J Sports Med Phys 2012; 52: 33-39.
[2] Faigenbaum AD, Myer GD. Resistance training among young athletes: safety, efficacy and injury prevention effects. Br J Sports Med 2010; 44: 56-63.
[3] Aginsky J, Neophytou N, Charalambous T. Isokinetic hamstring and quadriceps muscle strength profiles of elite South African football players. Afr J Phys Heal Educ Recreat Danc 2015; 20: 1225-1236.
[4] Greco CC, Da Silva WL, Camarda SRA, Denadai BS. Rapid hamstrings/quadriceps strength capacity in professional soccer players with different conventional isokinetic muscle strength ratios. J Sports Sci Med 2012; 11: 418-422.
[5] Jordan MJ, Aagaard P, Herzog W. Rapid hamstrings/quadriceps strength in ACL-reconstructed elite Alpine ski racers. Med Sci Sports Exerc 2015; 47: 109-119.
[6] Ford-Smith CD, Wyman JF, Elswick RK, Fernandez T. Reliability of stationary dynamometer muscle strength testing in community-dwelling older adults. Arch Phys Med Rehabil 2001; 82: 1128-1132.
[7] Clarke HH. Comparison of instruments for recording muscle strength. Res Quart Am Assoc Heal Phys Educ Recreat 2013; 25: 398-411.
[8] Montgomery LC, Douglass LW, Deuster PA. Reliability of an isokinetic test of muscle strength and endurance. J Orthop Sports Phys Ther 1989; 10: 315-322.
[9] Tate CM, Williams GN, Barrance PJ, Buchanan TS. Lower extremity muscle morphology in young athletes: an MRI-based analysis. Med Sci Sports Exerc 2006; 38: 122-128.
[10] Manca A, Solinas G, Dragone D, Deriu F. Isokinetic testing of muscle performance: new concepts for strength assessment. Isokinet Exerc Sci 2015; 23: 69-75.
[11] Harman E. Principles of test selection and administration. In: Haff GG, Triplett NT, editors. Essentials of Strength Training and Conditioning. 4th ed. Champaign, IL, USA: Human Kinetics, 2008. pp. 238-247.
[12] Maayah MF, Al-Jarrah MD, El Zahrani SS, Alzahrani AH, Ahmedv ET, Abdelaziem AA, Lakshmanan G, Almawajdeh NA, Alsufiany MB, Asi YOM. Test-retest strength reliability of the Electronic Push/Pull Dynamometer (EPPD) in the measurement of the quadriceps and hamstring muscles on a new chair. Open J Intern Med 2012; 2: 123-128.
[13] Barber-Westin SD, Noyes FR, Galloway M. Jump-land characteristics and muscle strength development in young athletes: a gender comparison of 1140 athletes 9 to 17 years of age. Am J Sports Med 2006; 34: 375-384.
[14] Ferro E. Intra- and inter-reliability of lower extremity muscle strength measurements using a hand-held dynamometer with and without a stabilization strap. Int J Exerc Sci 2011; 2: 23.
[15] Kılın¸c BE, Kara A, Camur S, Oc Y, Celik H. Isokinetic dynamometer evaluation of the effects of early thigh diameter difference on thigh muscle strength in patients undergoing anterior cruciate ligament reconstruction with hamstring tendon graft. J Exerc Rehabil 2015; 11: 95-100.
[16] Lemmer JT, Hurlbut DE, Martel GF, Tracy BL, Ivey FM, Metter EJ, Fozard JL, Fleg JL, Hurley BF. Age and gender responses to strength training and detraining. Med Sci Sports Exerc 2000; 32: 1505-1512.
[17] Lindle RS, Metter EJ, Lynch NA, Fleg JL, Fozard JL, Tobin J, Roy TA, Hurley BF. Age and gender comparisons of muscle strength in 654 women and men aged 2093 yr. J Appl Physiol 1997; 83: 1581-1587.
[18] Derviˇsevi´c E, Hadˇzi´c V. Quadriceps and hamstrings strength in team sports: Basketball, football and volleyball. Isokinet Exerc Sci 2012; 20: 293-300.
[19] Akay MF, Abut F, Oz¸cilo˘glu M, Heil D. Identifying the discriminative predictors of upper body power of cross- ¨ country skiers using support vector machines combined with feature selection. Neural Comput Appl 2016; 27: 1785-1796.
[20] Abut F, Akay MF. Machine learning and statistical methods for the prediction of maximal oxygen uptake: recent advances. Med Device 2015; 8: 369-379.
[21] Acikkar M, Akay MF, Ozgunen KT, Aydin K, Kurdak SS. Support vector machines for aerobic fitness prediction of athletes. Expert Syst Appl 2009; 36: 3596-3602.
[22] Vapnik V. The Nature of Statistical Learning Theory. 2nd ed. New York, NY, USA: Springer, 2000.
[23] Hsu CW, Chang CC, Lin CJ. A Practical Guide to support Vector Classification. Taipei, Taiwan: National Taiwan University, 2003.
[24] Mattera D, Haykin S. Support vector machines for dynamic reconstruction of a chaotic system. In: Sch¨olkopf B, Burges C, editors. Advances in Kernel Methods. Cambridge, MA, USA: MIT Press, 1999. pp. 211-241.
[25] Taheri SM, Hesamian G. A generalization of the Wilcoxon signed-rank test and its applications. Stat Pap 2012; 54: 457-470.
[26] Anaene Oyeka IC, Ebuh GU. Modified Wilcoxon signed-rank test. Open J Stat 2012; 2: 172-176.
[27] Tan S, Wang J, Liu S. Establishment of the prediction equations of 1RM skeletal muscle strength in 60- to 75-yearold Chinese men and women. J Aging Phys Act 2015; 23: 640-646.
[28] Muraki S, Fukumoto K, Fukuda O. Prediction of the muscle strength by the muscle thickness and hardness using ultrasound muscle hardness meter. Springerplus 2013; 2: 457.
[29] Maeda T, Oowatashi A, Kiyama R, Yoshida Y, Sakae K. Prediction of muscle strength using length and width of the bone. J Phys Ther Sci 2001; 13: 27-30.
[30] Mannion AF, Adams MA, Cooper RG, Dolan P. Prediction of maximal back muscle strength from indices of body mass and fat-free body mass. Rheumatology 1999; 38: 652-655.
[31] McNair PJ, Colvin M, Reid D. Predicting maximal strength of quadriceps from submaximal performance in individuals with knee joint osteoarthritis. Arthritis Care Res (Hoboken) 2011; 63: 216-222.