Modeling the Relationship between Training and Performance - A Comparison of Two Antagonistic Concepts

Few attempts have been made to apply systems theory to the description of human responses during physical training. Initially, Calvert, Banister, Savage & Bach (1976) proposed describing systems behavior with two antagonistic transfer functions ascribed to fitness as a positive and fatigue as a negative response to physical training. Performance, i.e. system output, was thus the balance between fitness and the fatigue effects calculated by a system of differential equations. This approach has been used in several studies to model the relationship between training and performance, but recently some authors have criticized the FF-Model for its methodical limitations and inconsistent empirical findings. Largely decoupled from this discussion another antagonistic model has been developed by Perl (2002). In order to analyze and optimize physiological adaptation processes, the so-called PerformancePotential-Model (PerPot) helps to simulate the interaction between training load and performance by using a dynamical stateevent-model with adaptive delay in effect. To compare these two antagonistic models with regard to some critical considerations two training studies (untrained subjects) on a cycle ergometer were carried out. The results show, that in nine out of fifteen cases, better model fit to real performance data is achieved with PerPot. The prediction of the performance values for the final two weeks of the training experiment were, indeed, on average of higher quality for PerPot. But regarding to the individual cases with the FF-Model, prediction of values succeeds to a smaller middle percentage deviation in eight of the fifteen subjects. Furthermore, in both models a better model-fit and prediction accuracy was achieved by equidistant time interval between the training and testing sessions.