Modelling in physiology and human performance : the influence of body size.
This thesis examined the validity of allometric models (Y = aXbg) in scaling physiological
and human performance data (Y) for differences in body size (X). 1) Anaerobic
performance. External peak power output (PPO) derived from supramaximalleg ergometry
was compared in young adult males and females, using a multivariate allometric model.
Estimated fat free mass (FFM) and thigh muscle-and-bone cross-sectional area served as
indicators of involved musculature. Male PPO was greater than female (P < 0.05), after
allometric adjustment for body size differences. This finding is questionable, however, as
the within-gender goodness-of-fit values for the regression models were very poor.
2) Cardiac dimensions. The proper relationships between echocardiographic dimensions
[left ventricular (LV) mass, and LV internal dimensions] and various indicators of overall
body size [height, body mass (BM), FFM, and body surface area (BSA)] were examined in
young, apparently healthy, adult males and females. Scaling by FFM was associated with the
least residual error in these samples. The obtained relationships were generally
dimensionally consistent, that is, LV mass proportional to FFM to the first power, and LV
internal dimensions related to the 1/3 power ofFFM. 3) Methodological issues. The
multivariate allometric scaling of peak oxygen uptake by height and BM was investigated.
Regression diagnostics revealed that the obtained exponents were unstable, and potentially
numerically inaccurate, due to severe collinearity between height and BM in the sample. For
elite weightlifting performance, detailed examination of the allometric regression residuals
revealed that the model was poorly specified. Re-specification of the model using secondorder
polynomials provided the optimal scaling of this data set.