Objective: Dynamical analysis can be used to study the changes of self-regulated biological processes driven by external stimuli. Recently, heart rate during effort tests has been successfully adjusted with a simple first order differential equation with constant coefficients driven by the body power expenditure. Although producing proper estimations and yielding pertinent indices to analyze such measurements, this approach suffers from its inability to model the saturation of the heart rate increase at high power expenditure and the change of heart rate equilibrium after effort. Approach: We propose a new analysis allowing to estimate the changes of heart rate response to effort (gain) as a function of the power expenditure value. Main results: When applied to heart rate measured on 30 amateur athletes performing a maximum graded treadmill effort test, the proposed model was able to predict 99% of the measured heart rate change during exercise. The gains estimated decreased with power increase above the first ventilatory threshold. This trend was stronger above the second ventilatory threshold and was strongly correlated with the maximum oxygen consumption. Significance: The proposed approach yields a highly precise modeling of heart rate dynamics during variable effort reflecting the changes of metabolic energy systems at play during exercise.