Psychophysiological dynamics of emotional reactivity: Interindividual reactivity characterization and prediction by a machine learning approach.

Abstract The fast reaction of the autonomic nervous system (ANS) to an emotional challenge (EC) is the result of a functional coupling between parasympathetic (PNS) and sympathetic (SNS) branches. This coupling can be characterized by measures of cross-correlations between electrodermal activity (EDA) (under the influence of the SNS) and the RR interval (the interval between R peaks) (under the influence of the PNS and the SNS). Significant interindividual variability has previously been reported in SNS-PNS coupling in emotional situations, and the present study aimed to identify interindividual cross-correlation variability in ANS reactivity. We therefore studied EDA and the RR interval in 62 healthy subjects, recorded during a 24-minute EC. A Gaussian Mixture Model was used to cluster tonic EDA-RR cross-correlations during the EC. This identified two clusters that were characterized by significant or non-significant cross-correlations (SCC and NCC clusters, respectively). The SCC cluster reported higher negative emotion after the EC, while the NCC cluster reported higher scores on the Center for Epidemiologic Studies–Depression scale. The latter finding suggests that NCC is a pathological mood pattern with altered negative perception. Furthermore, a machine learning model that included three parameters indexing the functionality of both branches of the ANS, measured at baseline, predicted cluster membership. Our results are a first step in detecting dysfunctional ANS reactivity in general population.