Using Critical Power To Predict Ramp Incremental Cycling Performance: Three Parameters Are Better Than Two

Abstract

The two-parameter critical power (CP2p) model states that the tolerable duration (Tlim) of severe-intensity cycling is determined by critical power (CP) itself, and the finite energy store that may be expended at work rates above CP (W′). Notwithstanding its ability to provide useful predictions of Tlim across a variety of exercise modes, the two-parameter model (CP2p) consistently overestimates Tlim for ramp incremental exercise. PURPOSE: To determine whether a three-parameter model of CP (CP3p) provides more accurate predictions of Tlim compared with those made by the CP2p model. METHODS: Seventeen healthy, recreationally-active adults (1 female; age: 29 ± 4 yrs, BMI: 25 ± 3, peak O2 uptake: 50 ± 8 ml·kg-1·min) completed a ramp cycling protocol, and a series of exhaustive, constant work rate (CWR) trials across 5 separate visits (~70-100% peak work rate). The CWR trials were used to establish each participant’s power-Tlim relationship, from which the CP2p and CP3p models were fitted to data. Cross-validation (CV) was used to assess external model validity. Finally, the observed Tlim during the ramp incremental protocol was compared to that predicted by the CP2p and CP3p models. RESULTS: CP2p was higher than CP3p (239 ± 14 W v 233 ± 13 W, P < 0.05), whereas W′ was smaller for the two- v three-parameter model (20.3 ± 1.3 kJ v 27.2 ± 2.9 kJ, P < 0.05). The CP3p model yielded a better fit to power-Tlim data than the CP2p model, as judged by the lower root-mean-square error (RMSE) computed from the CV procedure (128 s v 141 s). The predicted ramp Tlim obtained via the CP2p model was longer than the actual Tlim for ramp incremental cycling (∆21.9 s, P < 0.05); however, the ramp Tlim predicted by the CP3p model was not different from the actual ramp Tlim (∆1.7 s, P < 0.05). Furthermore, there was higher absolute agreement between actual and predicted ramp Tlim for the CP3p compared with the CP2p model, as evidenced by a higher concordance correlation coefficient (0.98 v 0.94) and lower RMSE (16.4 s v 27.7 s). CONCLUSIONS: Our findings indicate that the CP3p model provides better predictions of ramp exercise performance than the CP2p model. These findings provide further support for the idea that Tlim for supra-CP cycling is determined not only by the magnitude of W′, but also by a maximal rate at which W′ can be accessed, particularly at high work rates.