Estimation of Skin Conductance Response Through Adaptive Filtering

The importance of medical wearable sensors is increasing in aiding both diagnostic and therapeutic protocols, in a wide area of health applications. Among them, the acquisition and analysis of electrodermal activity (EDA) may help in detecting seizures and different human emotional states. Nonnegative deconvolution represents an important step needed for decomposing the measured galvanic skin response (GSR) in its tonic and phasic components. In particular, the phasic component, also known as skin conductance response (SCR), is related to the sympathetic nervous system (SNS) activity, since it can be modeled as the linear convolution between the SCR driver events, modeled by sparse impulse signals, with an impulse response representing the sudomotor SNS innervation. In this paper, we propose a novel method for implementing this deconvolution by an adaptive filter, determined by solving a linear prediction problem, which results independent on the impulse response parameters, usually represented by sampling the biexponential Bateman function. The performance of the proposed approach is evaluated by using both synthetic and experimental data.