Outputs from a fibre-optic skin-surface sensor have been mathematicall
y or statistically investigated for better extraction of information a
bout the skin-surface conditions. Methods employed in this study inclu
de simple or multiple regression analysis, functional fitting (for ext
racting characteristic parameters) and principal-component analysis. T
he regression analysis using the sensor outputs and the skin water con
tent has made it clear that a large applied force of the sensor is pre
ferable and that a measurement time period of 6-7 s is required to obt
ain the maximum correlation. For reduction of the measurement time per
iod, the use of multiple correlation has been found to be advantageous
. Principal-component analysis using the sensor outputs has revealed t
hat the first principal component has a much larger contribution ratio
and corresponds to the degree of skin hydration. However, it is not c
lear what the second component means.