Improved estimation of sweating based on electrical properties of skin

Abstract

Skin conductance (SC) has previously been reported to correlate strongly with sweat rate (Swr) within subjects, but weakly between subjects. Using a new solution for simultaneous recording of SC, skin susceptance (SS) and skin potential (SP) at the same skin site, the aim of this study was to assess how accurately sweat production can be estimated based on combining these electrical properties of skin. In 40 subjects, SC, SS, SP and Swr by skin water loss was measured during relaxation and mental stress. SC and Swr had high intraindividual correlations (median r = 0.77). Stepwise multilinear regression with bootstrap validation lead to a sweating estimation model based on the sum of SC increases, the SP area under the curve and the SS area under the curve, yielding an interindividual accuracy of R(2) = 0.73, rmse = 12.9%, limits of agreement of +27.6, -30.4% and an intraclass correlation coefficient of 0.84. Bootstrapping of training and test-sets gave median rmse = 15.4%, median R(2) = 0.66. The model was also validated for intraindividual variability. The results show that estimation of sweating is significantly improved by the addition of SS and SP measurement.