Development and validation of an unsupervised scoring system (Autonomate) for skin conductance response analysis

The skin conductance response (SCR) is increasingly being used as a measure of sympathetic activation concurrent with neuroscience measurements. We present a method of automated analysis of SCR data in the contexts of event-related cognitive tasks and nonspecific responding to complex stimuli. The primary goal of the method is to accurately measure the classical trough-to-peak amplitude of SCR in a fashion closely matching manual scoring. To validate the effectiveness of the method in event-related paradigms, three archived datasets were analyzed by two manual raters, the fully-automated method (Autonomate), and three alternative software packages. Further, the ability of the method to score non-specific responses to complex stimuli was validated against manual scoring. Results indicate high concordance between fully-automated and computer-assisted manual scoring methods. Given that manual scoring is error prone, subject to bias, and time consuming, the automated method may increase the efficiency and accuracy of SCR data analysis.