Edinburgh Napier University

  Skin temperature is a fundamental variable in human thermo-physiology, and yet skin temperature measurement remains impractical in most free-living, exercise and clinical settings, using currently available hard-wired methods. The purpose of this study was to compare wireless iButtons® and hard-wired thermistors for human skin temperature measurement. In the first of two investigations, iButtons® and thermistors monitored temperature in a controlled water bath (range: 10–40 °C) and were referenced against a certified, mercury thermometer. In the second investigation, eight healthy males completed three randomized trials (ambient temperature = 10 °C, 20 °C and 30 °C) while both devices recorded skin temperature at rest (in low and high wind velocities) and during cycle-ergometry exercise. The results are as follows; Investigation 1: both devices displayed very high validity correlation with the reference thermometer (r > 0.999). Prior to correction, the mean bias was +0.121 °C for iButtons® and +0.045 °C for thermistors. Upon calibration correction the mean bias for iButtons® and thermistors was not significantly different from zero bias. Interestingly, the typical error of the estimate of iButtons® (0.043 °C) was 1.5 times less than that of thermistors (0.062 °C), demonstrating iButtons'® lower random error. Investigation 2: the offset between iButton® and thermistor readings was generally consistent across conditions; however, thermistor responses gave readings that were always closer to ambient temperature than those given by iButtons®, suggesting potential thermistor drift towards environmental conditions. Mean temperature differences between iButtons® and thermistors during resting trials ranged from 0.261 °C to 1.356 °C. Mean temperature differences between iButtons® and thermistors during exercise were 0.989 °C (ambient temperature = 10 °C), 0.415 °C (ambient temperature = 20 °C) and 0.318 °C (ambient temperature = 30 °C). Observed error estimates were within the acceptable limits for the skin temperature method comparison, with typical errors <0.3 °C, correlation coefficients >0.9 and CV <1% under all conditions. These findings indicate that wireless iButtons® provide a valid alternative for human skin temperature measurement during laboratory and field investigations particularly when skin temperature measurement using other currently available methods may prove problematic.