Predictive validity of a computer model of body temperature during exercise.

The predictive validity of a computer model of human temperature regulation is tested by comparison with experimental data. Three male subjects were exercised at five different rates (B.M.R., 100 W, 150 W, 200 W, and 250 W) on a cycle ergometer in a controlled-environmental facility. Thermal conditions ranged from 13 degrees to 29 degrees C on the Effective Temperature (E.T.) scale. Two core (rectal and tympanic) and four skin temperature sites (2 torso, 1 leg, and 1 arm) were monitored. Experimental figures for core temperature (TC) and mean body temperature (MBT), evolved from the six monitored sites, were compared with matched simulation data from the computer model. A high negative correlation (r = -0.87) was found for increasing "effective temperature" and mean absolute difference (d) between experimental and simulation data for mean body temperatures. The model has increasing predictive validity as higher heat stress is encountered (E.T. greater than 25 degrees C, d MBT less than 0.3 degrees C), which decreases (E.T. less than 16 degrees C, d MBT greater than 0.8 degrees C) in reduced environmental temperatures.