Conditional entropy approach to analyze cognitive dynamics in autism spectrum disorder

ABSTRACT

Objective

Preliminary evidence has documented functional connectivity during the cognitive task in Autism Spectrum Disorder (ASD). However, evidence of effective neural connectivity with respect to information flow between different brain regions during complex tasks is missing. The present paper aims to provide insights into the cognition-based neural dynamics reflecting information exchange in brain network under cognitive load in ASD.

Methods

Twenty-two individuals with ASD (8–18 years) and 18 Typically Developing (TD; 6–17 years) individuals participated in the cognitive task of differentiating risky from neutral stimuli. The Conditional Entropy (CE) technique is applied upon task-activated Electroencephalogram (EEG) to measure the causal influence of the activity of brain’s one Region of interest (ROI) over another.

Results

A higher CE in frontal ROI and left hemisphere reflected atypical brain complexity in ASD. The absence of causal effect, poor Coupling Strength (CS; measured using CE) and hemisphere lateralization is responsible for lower cognition in ASD. However, the persistent information exchange during the task reflects the existence of certain alternative paths when other direct paths remained disconnected due to cognitive impairment. The Support Vector Machine (SVM) classifier showed that CE can identify the atypical information exchange with an accuracy of 96.89% and area under curve = 0.987.

Discussion

The statistical results reflect a significant change in the information flow between different ROIs in ASD. A correlation of CS and behavioral domain suggests that the cognitive decline could be predicted from the connectivity patterns. Thus, CS could be a potential biomarker to identify cognitive status at a higher discrimination rate in ASD.

KEYWORDS:

• Autism spectrum disorder
• causality
• conditional entropy
• coupling
• information flow
• support vector machine

Disclosure statement

The authors declare no conflict of interest.

Supplementary material

Supplemental data for this article can be accessed here.

Additional information

Funding

The present work is not supported financially by any funding agency.

Notes on contributors

Tanu Wadhera

Tanu Wadhera received the B.tech degree in Electronics and Communication from Guru Nanak Engineering College, Ludhiana, India and M.tech degree in Eletronics & Communication from Punjabi University, Patiala, India. Currently, she is a Research Scholar in National institute of Technology, Jalandhar, India.  Her research interests are cognitive neuroscience, behavioral understanding, technology in Autism Spectrum Disorder, biomedical signal processing, signal classification and prediction of neurodevelopmental disorders.

Deepti Kakkar

Deepti Kakkar did her Bachelor of Technology in Electronics and Communication Engineering from Himachal Pradesh University, India in 2003 and Masters of Engineering in electronics product design and technology from Punjab University, Chandigarh. Deepti did her PhD in Cognitive Radios from Dr. B.R. Ambedkar National Institute of Technology, Jalandhar, India. She has a total academic experience of 15 years and at present she is ASSISTANT PROFESSOR in Electronics and Communication department with Dr. B. R. Ambedkar National Institute of Technology, Jalandhar, India. Her recent research interests include cognitive neuroscience, neuro-developmental disorder, dynamic spectrum allocation, spectrum sensing, and Cognitive Radios.