Individual response to exercise training - a statistical perspective

J Appl Physiol (1985). 2015 Jun 15;118(12):1450-9. doi: 10.1152/japplphysiol.00714.2014. Epub 2015 Feb 5.


In the era of personalized medicine, interindividual differences in the magnitude of response to an exercise training program (subject-by-training interaction; "individual response") have received increasing scientific interest. However, standard approaches for quantification and prediction remain to be established, probably due to the specific considerations associated with interactive effects, in particular on the individual level, compared with the prevailing investigation of main effects. Regarding the quantification of subject-by-training interaction in terms of variance components, confounding sources of variability have to be considered. Clearly, measurement error limits the accuracy of response estimates and thereby contributes to variation. This problem is of particular importance for analyses on the individual level, because a low signal-to-noise ratio may not be compensated by increasing sample size (1 case). Moreover, within-subject variation in training efficacy may contribute to gross response variability. This largely unstudied source of variation may not be disclosed by comparison to a control group but calls for repeated interventions. A second critical point concerns the prediction of response. There is little doubt that exercise training response is influenced by a multitude of determinants. Moreover, indications of interaction between influencing factors of training efficacy lead to the hypothesis that optimal predictive accuracy may be attained using an interactive rather than additive approach. Taken together, aiming at conclusive inference and optimal predictive accuracy in the investigation of subject-by-training interaction entails specific requirements that are deducibly based on statistical principles but beset with many practical difficulties. Therefore, pragmatic alternatives are warranted.

Keywords: determinant; interaction; moderator; prediction; variance components.

Publication types

  • Research Support, Non-U.S. Gov't
  • Review

MeSH terms

  • Data Interpretation, Statistical*
  • Exercise / physiology*
  • Humans
  • Physical Education and Training / statistics & numerical data*