Electrodermal signal analysis using continuous wavelet transform as a tool for quantification of sweat gland activity in diabetic kidney disease.

Sympathetic innervation of the sweat gland (SG) manifests itself electrically as electrodermal activity (EDA), which can be utilized to measure sudomotor function. Since SG exhibits similarities in structure and function with kidneys, quantification of SG activity is attempted through EDA signals. A methodology is developed with electrical stimulation, sampling frequency and signal processing algorithm. One hundred twenty volunteers participated in this study belonging to controls, diabetes, diabetic nephropathy, and diabetic neuropathy. The magnitude and time duration of stimuli is arrived by trial and error in such a way it does not influence controls but triggers SG activity in other Groups. This methodology leads to a distinct EDA signal pattern with changes in frequency and amplitude. The continuous wavelet transform depicts a scalogram to retrieve this information. Further, to distinguish between Groups, time average spectrums are plotted and mean relative energy (MRE) is computed. Results demonstrate high energy value in controls, and it gradually decreases in other Groups indicating a decline in SG activity on diabetes prognosis. The correlation for the acquired results was determined to be 0.99 when compared to the standard lab procedure. Furthermore, Cohen's value, which is less than 0.25 for all Groups indicating the minimal effect size. Hence the obtained result is validated and statistically analyzed for individual variations. Thus this has the potential to get transformed into a device and could prevent diabetic kidney disease.