Identifying Task Specific Subgroups in Autism Spectrum Disorder (ASD)

Background: Atypical autonomic nervous system (ANS) activations are commonly observed in individuals with autism spectrum disorder (ASD). These atypical activations are often manifested in irregularities in cardiac activities and the function of sweat glands.  

Objectives: The goal of this study is to identify subgroups of individuals with ASD by analyzing their physiological signals in order to discriminate between specific tasks. The tasks address one of the affected domains of ASD which is anxiety. 

Methods: 40 children with ASD participated in this study. We measured electrocardiogram (ECG) and electrodermal (EDA) activities from participants while they performed a sequence of different tasks including: baseline, stroop task, baseline, public speaking task, and baseline. The task in baseline was watching video. Public speaking and stroop tasks were performed in order to elicit anxiety.

Results: ECG and EDA observations were represented in terms of various time and frequency meta-features. Feature clustering with different parameters and number of clusters was then performed to examine the pair-wise separability of the tasks for the ASD group. A weak statistical correspondence between the cluster membership of the resulting clustering and the pairs of tasks under investigation was found in all pair-wise cases.

Conclusions: The lack of separability between pairs of tasks, evaluated in this work, might indicate that ASD population demonstrates non-discriminative electrocardiogram and electrodermal activations while engaged in these tasks. Other physiological observations (e.g. respiratory) might be needed to separate the tasks.