Modeling the sweat regulation mechanism is important for reliable simulation of the human thermoregulatory processes. The complexity of the mechanism makes it very difficult to model using traditional techniques. An engineering or systems overview of the human thermoregulatory system is reported. An extensive review of previous attempts to model the human sweat rate forms an important part of this paper.
In addition, this study investigates the applicability of neural networks to the problem of modeling the complex nonlinearities of the sweat regulatory mechanism. It is believed that neural networks provide better generalization capabilities for all the cited dependencies resulting in better sweat prediction models. The network is thus in a position to generalize based on the different operating conditions and provide more reliable outputs over an entire range of environments and metabolic profiles. Results of a neural network sweat rate model for a sedentary human exposed to step changes in environmental temperature are reported.
