Estimation of lactate threshold with machine learning techniques in recreational runners

Abstract

Lactate threshold is considered an essential parameter when assessing performance of elite and recreational runners and prescribing training intensities in endurance sports. However, the measurement of blood lactate concentration requires expensive equipment and the extraction of blood samples, which are inconvenient for frequent monitoring. Furthermore, most recreational runners do not have access to routine assessment of their physical fitness by the aforementioned equipment so they are not able to calculate the lactate threshold without resorting to an expensive and specialized center. Therefore, the main objective of this study is to create an intelligent system capable of estimating the lactate threshold of recreational athletes participating in endurance running sports.The solution here proposed is based on a machine learning system which models the lactate evolution using recurrent neural networks and includes the proposal of standardization of the temporal axis as well as a modification of the stratified sampling method. The results show that the proposed system accurately estimates the lactate threshold of 89.52% of the athletes and its correlation with the experimentally measured lactate threshold is very high (R=0,89). Moreover, its behaviour with the test dataset is as good as with the training set, meaning that the generalization power of the model is high.Therefore, in this study a machine learning based system is proposed as alternative to the traditional invasive lactate threshold measurement tests for recreational runners.

A heuristic approach for lactate threshold estimation for training decision-making: An accessible and easy to use solution for recreational runners

“…Figure 1 represents this concept. In [8], we proposed an individual acceptable error for recreational athletes which we represent in table 1. However, the evaluation of the system accuracy for an estimator created from empirical data is not straightforward.…”

“…Therefore, to test the limits of the initial estimator and consequently the learning, performance evaluation and estimator selection methodologies used in [8], new experimental tests were made with 50 recreational athletes. The purpose of these experimental data is to serve as a test set to evaluate the initial estimator, assess its bias-variance performance and adjust the estimator accordingly in this 2 nd iteration.…”

“…In this work, a feature transformation is used to represent the LT in a way that it is decoupled from the PTS, so that it can be represented in relative terms whit respect to the individual endurance running speed reserve (5) As we highlighted in [8,14], there is a consistent association between the PTS and the LT. As shown in figure 6, once we dis-attach the LT from the effect of the PTS, the variability of the LT is highly reduced and the problem is simplified. Then, using the % of endurance running speed reserve corresponding to the LT as estimand, we propose to use the mean as an extremely parsimonious estimator.…”

“…However, non of the proposed methods were able to achieve both a reasonable reliability while addressing accessibility and facilitating its consistent use. To the best of our knowledge, the model we previously proposed as a LT estimator [8] is the only alternative that has covered all these operational requirements. However, creating an estimator from empirical data is an iterative process of problem discovery where the complexity of the problem and its relation with the available data is further discovered with each iteration.…”

“…In [8] we run a first iteration and proposed a first model as estimator, hereafter referred as the initial estimator. The empirical LT was obtained using the Dmax LT protocol, which is the most recommended method for the LT estimation from blood samples [3].…”