Relationships Between Model Estimates and Actual Match-Performance Indices in Professional Australian Footballers During an In-Season Macrocycle

摘要

Purpose: To assess and compare the validity of internal and external Australian football (AF) training-load measures for predicting match exercise intensity (MEI/min) and player-rank score (PRScore) using a variable dose-response model. Methods: A cohort of 25 professional AF players (23 +/- 3 y, 188.3 +/- 7.2 cm, 87.7 +/- 8.4 kg) completed a 24-wk in-season macrocycle. In-season internal training and match load were quantified using session rating of perceived exertion (sRPE) and external load from satellite and accelerometer data. Using a training-impulse (TRIMP) calculation, external load (au) was represented as distance (TRIMPDist), distance = 4.16 m/s (TRIMPHSDist), and PlayerLoad (TRIMPPL). In-season training load, MEI/min, and PRScore were applied to a variable dose-response model, which provided estimates of MEI/min and PRScore. Predicted MEI/min and PRScore were correlated with actual measures from each match. The magnitude of the difference between MEI/min and PRScore estimates for each model input and the difference between the precision of internal and external load measures to predict MEI/min and PRScore were calculated using the effect size +/- 90% confidence interval (CI). Results: Estimates of MEI/min demonstrated very large associations with actual MEI/min (r, 90% CI) (eg, TRIMPDist.76 +/- .13, and sRPE(Skills) .73 +/- .14). Estimates of PRScore demonstrated associations of large magnitude with actual PRScore using the same inputs. Precision of actual MEI/min was lowest using sRPE compared with (ES +/- 90% CI) TRIMPDist, -.67 +/- .34, and TRIMPPL, -.91 +/- .39. There were trivial and unclear differences in the precision of PRScore estimates between TRIMP and sRPE inputs. Conclusions: Dose-response models from multiple training-load inputs can predict within-individual variation of MEI/min and PRScore. Internal and external training-input methods exhibited comparable predictive power.