Machine learning and statistical methods for the prediction of maximal oxygen uptake: recent advances

Abut, Fatih; Akay, Mehmet Fatih

doi:10.2147/mder.s57281

Cited by 23 publications

(14 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Predictor Due to fact that SVM [18], in the literature, has been often reported to be superior to other machine learning methods especially in the field of sport physiology [1,19,20], it has been chosen to build the VO 2 max prediction models. SVM constructs a hyperplane or set of hyperplanes in a high-or infinite-dimensional space to perform a regression analysis.…”

Section: Modelsmentioning

confidence: 99%

A robust ensemble feature selector based on rank aggregation for developing new VO2max prediction models using support vector machines

Abut¹,

Akay²,

George³

2019

Turk J Elec Eng & Comp Sci

Self Cite

View full text Add to dashboard Cite

This paper proposes a new ensemble feature selector, called the majority voting feature selector (MVFS), for developing new maximal oxygen uptake (VO2max) prediction models using a support vector machine (SVM). The approach is based on rank aggregation, which meaningfully utilizes the correlation among the relevance ranks of predictor variables given by three state-of-the-art feature selectors: Relief-F, minimum redundancy maximum relevance (mRMR), and maximum likelihood feature selection (MLFS). By applying the SVM combined with MVFS on a self-created dataset containing maximal and submaximal exercise data from 185 college students, several new hybrid VO2max prediction models have been created. To compare the performance of the proposed ensemble approach on prediction of VO2max, SVM-based models with individual combinations of Relief-F, mRMR, and MLFS as well as with other alternative ensemble feature selectors from the literature have also been developed. The results reveal that MVFS outperforms other individual and ensemble feature selectors and yields up to 8.76% increment and 11.15% decrement rates in multiple correlation coefficients (Rs) and root mean square errors (RMSEs), respectively. Furthermore, in addition to reconfirming the relevance of sex, age, and maximal heart rate in predicting VO2max, which were previously reported in the literature, it is revealed that submaximal heart rates and exercise times at 1.5-mile distance are two further discriminative predictors of VO2max. The results have also been compared to those obtained by a general regression neural network and single decision tree combined with MVFS, and it is shown that the SVM exhibits much better performance than other methods for prediction of VO2max.

show abstract

Section: Modelsmentioning

confidence: 99%

A robust ensemble feature selector based on rank aggregation for developing new VO2max prediction models using support vector machines

Abut¹,

Akay²,

George³

2019

Turk J Elec Eng & Comp Sci

Self Cite

View full text Add to dashboard Cite

show abstract

“…When it comes to comparing linear and nonlinear models with regard to prediction accuracy, the quality of prediction of neural models seems to be similar to that of regression analysis and regression models, potentially even better. This seems to hold true in VO 2 max prediction (Abut, & Akay, 2015;Akay, Zayid, Aktürk, & George, 2011) as well as in performance prediction studies (Edelmann-Nusser, Hohmann, & Henneberg, 2002;Maszczyk, Zajac, & Ryguła, 2011;Maszczyk, Rocznoik, Waskiewicz, Czuba, Mikolajec, Zajac, & Stanula, 2012). Hence, Jäger, Kurz and Müller (2016) assume that nonlinear neural network approaches possibly provide more sophisticated methods for predicting maximal mean speed in a 4x1000 m Field Test which may enhance the accuracy of their linear model.…”

Section: Introductionmentioning

confidence: 97%

“…Besides those studies which measure VO 2 max directly in a maximal incremental exercise test on a treadmill or cycle ergometer, there are also some approaches that try to predict VO 2 max based on non-exercise and submaximal exercise test data (e.g., Black, Vehrs, Fellingham, George & Hager, 2016) or even based on statistical models using solely non-exercise data (see Abut & Akay, 2015, for a review). In case of non-exercise models, variables such as sex, age or body mass index were used to predict VO 2 max.…”

Section: Introductionmentioning

confidence: 99%

Linear and Nonlinear Prediction Models Show Comparable Precision for Maximal Mean Speed in a 4x1000 m Field Test

Jäger

Kurz

Müller

2017

International Journal of Computer Science in Sport

View full text Add to dashboard Cite

Maximal oxygen uptake (VO 2 max) is one of the most distinguished parameters in endurance sports and plays an important role, for instance, in predicting endurance performance. Different models have been used to estimate VO 2 max or performance based on VO 2 max. These models can use linear or nonlinear approaches for modeling endurance performance. The aim of this study was to estimate VO 2 max in healthy adults based on the Queens College Step Test (QCST) as well as the Shuttle Run Test (SRT) and to use these values for linear and nonlinear models in order to predict the performance in a maximal 1000 m run (i.e. the speed in an incremental 4x1000 m Field Test (FT)). 53 female subjects participated in these three tests (QCST, SRT, FT). Maximal oxygen uptake values from QCST and SRT were used as (a) predictor variables in a multiple linear regression (MLR) model and as (b) input variables in a multilayer perceptron (MLP) after scaling in preprocessing. Model output was speed [km·h for MLP. Results showed that the accuracy of the applied MLP was comparable to the MLR, but did not outperform the linear approach.

show abstract

“…Using inputs with known outputs, it can be trained and then utilized to predict unknown outputs. A 2015 review by Akay et al identified a variety of studies that attempted to use ANNs or other forms of machine learning to predict VO 2 max from maximal, submaximal, and non-exercise data [3]. Additionally, many studies have used ANNs to predict energy expenditure from accelerometer data [15,28,40,41,45].…”

Section: Introductionmentioning

confidence: 99%

Predicting oxygen uptake responses during cycling at varied intensities using an artificial neural network

Borror

Mazzoleni

Coppock

et al. 2019

Biomedical Human Kinetics

View full text Add to dashboard Cite

SummaryStudy aim: Oxygen Uptake (VO2) is avaluable metric for the prescription of exercise intensity and the monitoring of training progress. However, VO2 is difficult to assess in anon-laboratory setting. Recently, an artificial neural network (ANN) was used to predict VO2 responses during aset walking protocol on the treadmill [9]. The purpose of the present study was to test the ability of an ANN to predict VO2 responses during cycling at self-selected intensities using Heart Rate (HR), time derivative of HR, power output, cadence, and body mass data.Material and methods: 12 moderately-active adult males (age: 21.1 ± 2.5 years) performed a50-minute bout of cycling at a variety of exercise intensities. VO2, HR, power output, and cadence were recorded throughout the test. An ANN was trained, validated and tested using the following inputs: HR, time derivative of HR, power output, cadence, and body mass. A twelve-fold hold-out cross validation was conducted to determine the accuracy of the model.Results: The ANN accurately predicted the experimental VO2 values throughout the test (R2 = 0.91 ± 0.04, SEE = 3.34 ± 1.07 mL/kg/min).Discussion: This preliminary study demonstrates the potential for using an ANN to predict VO2 responses during cycling at varied intensities using easily accessible inputs. The predictive accuracy is promising, especially considering the large range of intensities and long duration of exercise. Expansion of these methods could allow ageneral algorithm to be developed for a more diverse population, improving the feasibility of oxygen uptake assessment.

show abstract

Machine learning and statistical methods for the prediction of maximal oxygen uptake: recent advances

Cited by 23 publications

References 20 publications

A robust ensemble feature selector based on rank aggregation for developing new VO2max prediction models using support vector machines

A robust ensemble feature selector based on rank aggregation for developing new VO2max prediction models using support vector machines

Linear and Nonlinear Prediction Models Show Comparable Precision for Maximal Mean Speed in a 4x1000 m Field Test

Predicting oxygen uptake responses during cycling at varied intensities using an artificial neural network

Contact Info

Product

Resources

About