Accuracy of next-generation wireless skin temperature sensors during exercise-heat stress

Skin temperature is fundamental in characterising human thermoregulatory responses. Wired probes, although accurate, restrict movement and are impractical outside laboratory settings. The iButton (DS1922L; Maxim Integrated, USA) is a widely used wireless alternative but does not meet the precision recommended by ISO 9886 and permits only retrospective data retrieval. A recently developed wireless sensor, the eTemp Performance (BodyCAP, France), claims to address these limitations but has not been validated independently. Here, we compared the eTemp and iButton against a wired thermocouple reference (SST-1; Physitemp, USA) during rest, exercise and recovery in the heat [mean (SD): 35.0 (0.5)°C, 40.6 (1.5)% relative humidity]. Twenty-six adults (10 women) completed seated rest (15 min), cycling at fixed heat production (30 min) and passive recovery (15 min). Mean skin temperature was calculated using the Ramanathan four-site formula. Across all periods, the eTemp showed a small negative bias (mean difference, -0.04°C; 95% limits of agreement, -0.33°C to 0.25°C; concordance correlation coefficient, 0.986), whereas the iButton showed a small positive bias (+0.11°C; 95% limits of agreement, -0.23°C to 0.46°C; concordance correlation coefficient, 0.974). The eTemp met all a priori acceptability thresholds for mean skin temperature. The iButton met the acceptability threshold for mean bias in all periods and was statistically equivalent to the reference; however, the 95% confidence interval on the upper limit of agreement marginally exceeded 0.5°C during exercise, recovery and overall. These findings provide the first independent validation of the eTemp Performance sensor and support the use of both wireless devices for mean skin temperature measurement during exercise-heat stress, notwithstanding the known limitations in accuracy of the iButton against the ISO 9886 standard.