Comparing EMG amplitude patterns of responses during dynamic exercise: Polynomial vs log-transformed regression

The purposes of this study were to determine if (a) the log-transformed model can be applied to dynamic exercise and (b) the slope and y-intercept terms can provide additional information above and beyond the polynomial regression analyses. Eleven physically active individuals performed incremental cycle ergometry on a single occasion. Electromyographic electrodes were placed on the three superficial quadriceps muscles to record muscle activation during the exercise test. The patterns of responses for electromyographic amplitude vs power output were analyzed with polynomial and log-transformed regression models. The results of the polynomial regression for the composite data indicated that the best-fit model for the vastus lateralis muscle was linear (R²=0.648, P<0.0001), whereas the best-fit model for the rectus femoris (R²=0.346, P=0.013) and vastus medialis (R²=0.764, P=0.020) muscles was quadratic. One-way repeated measures analyses indicated no significant differences (P>0.05) across the three superficial quadriceps muscles for the slope and y-intercept terms. These findings suggest that the log-transformed model may be a more versatile statistical approach to examining neuromuscular responses during dynamic exercise.