A comparison of modelling procedures used to estimate the power-exhaustion time relationship

This study aimed to test the consistency of using the power required to elicit maximal oxygen uptake during incremental test (P (t)) to demarcate the range of power intensity in the modelling of the power-exhaustion time relationship. Different mathematical procedures were tested using data from ten subjects exercising on a cycle ergometer. After the determination of P (t) and the power at the ventilatory threshold, the subjects did six tests at constant power to exhaustion within 2-15 min. Estimates were obtained from a segmented model using two distinct equations of the anaerobic contribution to power below and above P (t), respectively. This model fit the overall data with a better adequacy than the simple hyperbolic model (standard error of 29.2 +/- 25.2 vs. 42.3 +/- 25.2 s). The power asymptotes were 225.7 +/- 27.3 W from the segmented model, 226.2 +/- 27.3 and 283.3 +/- 20.5 W from the simple model applied to data below and above P (t), respectively. The estimates from the segmented model were strongly correlated with their analogues from the simple model applied only to data below P (t) (R = 1.00 for power asymptote and curvature coefficient). They were not correlated with their analogues from the simple model applied only to data above P (t). These discrepancies between modelling procedures could arise from the method used to determine P (t) and the oversimplification of the oxygen uptake kinetics. These limitations could lead the segmented model to an overestimation of the anaerobic contribution which was around 15% of total energy expended at P (t).