Developing novel analytics for ambulatory electrodermal activity data

Electrodermal activity (EDA) data contains rich and valuable information about changes in the sympathetic nervous system on a second-by-second timescale, providing insight into affective changes that may not be observed by gross behavioral measurements or informant rating scales alone. While the majority of research employing EDA to date is collected in clinic or research settings, ambulatory EDA measures in the home can provide longitudinal information about behavioral and affective changes that may be more ecologically valid and/or capture important context-dependent variance over time. However, like all EDA measures — regardless of setting, population or equipment used — ambulatory EDA is influenced by non-autonomic factors (i.e., thermoregulation, metabolic demand, participant movement, electrical noise). Therefore, it is critical to implement rigorous, well-defined and replicable quality control procedures for selecting EDA data for analysis and only use statistical parameters that are physiologically meaningful.

To address these issues, Matthew Goodwin and his colleagues Ian Kleckner (University of Rochester Medical Center) and Rebecca Jones (Weill Cornell Medicine) have made significant progress: (1) developing an automated quality-control procedure that can be tailored to any EDA study in both clinic and home settings¹; (2) reducing EDA data to a set of key metrics and summary statistics to assess associations with other clinical outcome measures; (3) establishing the use of an analytical framework appropriate for multiperson, multiweek collection of EDA data; and (4) revealing preliminary associations between measures of ambulatory EDA and parent reports of child behavior. This work was supported, in part, by an earlier SFARI Research Award.

The current project will further establish an EDA analysis pipeline to include automated analysis of skin-conductance responses² in ambulatory and clinic data, develop and validate novel state-dependent EDA metrics to capture shifts in affective states over time, and assess the application of electrodermal lability in identifying trait-like versus state-like affective changes in a given person across contexts and time.

The knowledge and tools that Goodwin’s team plans to develop will ultimately be shared with the broader scientific community to accelerate the effective use of ambulatory EDA measures in clinical trials research involving individuals with autism and related neurodevelopmental disorders.