Non-invasive detection of the anaerobic threshold by a neural network model of the heart rate—work rate relationship

Abstract

The heart rates (HRs) indicating the anaerobic threshold, which corresponds to a fixed blood lactate concentration of 4mmol/l (heart rate at the onset of blood lactate accumulation (HR-OBLA)); are predicted on the basis of the heart rate—work rate relationship during exercise performance. The predictor is a multi-layer perceptron (MLP). For training and testing, 225 male soccer players (mean age, 21.6±4.5 years) performed an incremental running test on a treadmill with 3-min runs and 30-s blood sampling sessions for lactate assessment. In addition to the treadmill test data, the ages, masses, heights, body mass indices, and playing positions of half of the subjects (n=113) were used to train the MLP and the remaining 112 were used for external validation. The results showed that the HR values recorded in the last stages before exhaustion were strong predictors of HR-OBLA (r=0.875; standard error of estimates, 4.17). Inclusion of other parameters did not improve the prediction rate. Although additional work is needed to generalize the method to training prescription, results suggest that a neural network is a promising non-invasive tool for accurately estimating the HR-OBLA from exercise test data.