Elementary integrate-and-fire process underlies pulse amplitudes in Electrodermal activity.

Area of Science:

• Physiology
• Biophysics
• Statistical Modeling

Background:

• Electrodermal activity (EDA) reflects sweat-induced skin conductance changes.
• EDA pulsatile changes resemble an integrate-and-fire process.
• Previous work established inverse Gaussian distribution for EDA inter-pulse intervals.

Purpose of the Study:

• Characterize the statistical structure of EDA pulse amplitudes based on physiology.
• Hypothesize EDA pulse amplitude is proportional to excess sweat volume.
• Develop and test physiologically-based models for EDA amplitude.

Main Methods:

• Modeled EDA pulse amplitude as a difference of two inverse Gaussian distributions.
• Tested four simplifications of the proposed model.
• Compared models using data from healthy volunteers during wakefulness and sedation.
• Evaluated model performance using Akaike's Information Criterion (AIC) and quantile-quantile plots.

Main Results:

• Simplified inverse Gaussian models outperformed other tested models (lognormal, gamma).
• Model performance was consistent across wakeful and sedated cohorts.
• The best models provided a parsimonious and accurate description of EDA amplitude characteristics.

Conclusions:

• Physiologically-based probability models offer accurate descriptions of EDA temporal and amplitude dynamics.
• Simplified inverse Gaussian models provide a robust framework for EDA amplitude analysis.
• This work enables summarizing EDA amplitude information with a limited number of parameters.