Development of Novel Maximal Oxygen Uptake Prediction Models for Turkish College Students Using Machine Learning and Exercise Data


AKAY M. F., ÇETİN E., YARIM İ., Bozkurt O., ÖZÇİLOĞLU M. M.

9th International Conference on Computational Intelligence and Communication Networks (CICN), Girne, Cyprus (Kktc), 16 - 17 September 2017, pp.186-189 identifier

  • Publication Type: Conference Paper / Full Text
  • Volume:
  • Doi Number: 10.1109/cicn.2017.41
  • City: Girne
  • Country: Cyprus (Kktc)
  • Page Numbers: pp.186-189
  • Çukurova University Affiliated: Yes

Abstract

Maximal oxygen uptake (VO(2)max) is the maximum rate of oxygen consumption as measured during maximal exercise. The purpose of this study is to produce new prediction models for Turkish college students by using machine learning methods including Support Vector Machines (SVM), Generalized Regression Neural Networks (GRNN), Radial Basis Function Network (RBFN) and Decision Tree Forest (DTF). The dataset comprises data of 98 subjects and the predictor variables are gender, age, height, weight, maximum heart rate (HRmax), grade, speed and exercise time. Fifteen different VO(2)max prediction models have been created with the variables listed above. The performance of the prediction models has been calculated by using common metrics such as standard error of estimate (SEE) and multiple correlation coefficient (R). The results show that GRNN based models usually produced much lower SEE's and higher R's than the ones given by SVM, DTF and RBFN based models. On the other hand, the RBFN based models yielded the worst performance with unacceptable error rates. Also, this study shows that the predictor variables grade, speed and time play a significant role in VO(2)max prediction.