Laboratory and Home-Based Assessment of Electrodermal Activity in Individuals with ASD

Background:

Sweat secretion from the skin is modulated by postganglionic sympathetic fibers surrounding eccrine sweat glands. Because eccrine sweat glands receive only excitatory sympathetic nerve impulses, electrodermal activity (EDA) constitutes a purely sympathetic response. The most widely studied property of EDA is skin conductance, which can be quantified by applying electrical potential between two points of skin contact and measuring resulting current flow between them. Changes in EDA has been widely used as an objective peripheral index of cognition, arousal, emotion, and attention in the general population since it results from neuronal activity in the amygdala, premotor cortex, pre-frontal cortex, hypothalamus, reticular activating system, and hippocampus. However, because EDA recordings are also influenced by non-autonomic factors such as thermoregulation, metabolic demand, participant movement, electrical noise, and other artifacts, it is critical to implement rigorous, well-defined, and replicable quality control procedures for selecting data prior to conducting inferential statistical testing.

Objectives:

Our objectives were threefold. First, we sought to develop and evaluate a quality control procedure for automatically selecting EDA data for subsequent comparative testing. Second, after applying our procedure to identify valid segments of EDA data, we evaluated EDA responses across laboratory and in-home settings. Third, we conducted preliminary analyses on EDA reactivity and its association with a variety of experimental tasks in laboratory settings, over repeated measures in the home, and in relation to day-to-day fluctuations in child irritability as rated by parents using a smartphone.

Methods:

Twenty children and adolescents (5 females, range 5-13 yrs, M=8 yrs) with a confirmed diagnosis of ASD (ADOS calibrated severity score, M=7.9, SD=1.8) were observed repeatedly over an 8-week period in both laboratory and home settings while their EDA was wirelessly recorded with the Affectiva Q sensor and their day-to-day fluctuations in irritability was rated by parents using a smartphone.

Results:

First, starting with 181 hours of raw EDA data collected in the study, our automated quality control procedure identified 120 hours of valid data across the 20 participants, yielding, on average, 6 hours of quality data per participant across an average of 4.6 recording days. Second, EDA responses discriminated between a variety of negative vs neutral affect-inducing tasks in the laboratory (Brief Observation of Social Communication Change, Purdue Pegboard, passive movie watching, and an iPad puzzle). Third, patterns of EDA data identified individual differences within and between children with ASD assessed over 7-10 days in home settings based on tonic level, variability, and the ratio between sympathetic reactivity (increasing slope) and recovery (decreasing slope). Fourth, using Hierarchical Linear Modeling, mean and standard deviation EDA was found to predict daily ratings of irritability at p=0.002 and p=0.043, respectively.

Conclusions:

Being able to reliably and validly obtain peripheral measures of autonomic arousal from individuals with ASD in an untethered way in both laboratory and natural environments over time enables an efficient and scalable way to gather biomarker data that could be used in diagnostic, phenotypic, and clinical/intervention explorations in autism and related neurodevelopmental disorders.