EEG decoding for effects of visual joint attention training on ASD patients with interpretable and lightweight...

Area of Science:

• Neuroscience
• Cognitive Science
• Developmental Psychology

Background:

• Visual joint attention is crucial for social and language development in humans.
• Individuals with autism spectrum disorder (ASD) often exhibit deficits in visual joint attention.
• EEGNet is an effective model for decoding electroencephalogram (EEG) signals, but its application in ASD attentional training is underexplored.

Purpose of the Study:

• To investigate the effectiveness of joint attention training in improving visual attention in individuals with ASD.
• To utilize EEGNet and saliency mapping to decode P300 signals and visualize cognitive changes during visual attention tasks in ASD patients.

Main Methods:

• EEGNet was employed to decode P300 signals elicited during a joint attention training paradigm.
• Saliency map analysis was used to visualize the spatial and temporal patterns of brain activity.
• P300 signal characteristics, including latency and classification contribution, were analyzed before and after training in social and non-social scenarios.

Main Results:

• The parietal lobe, particularly the Pz electrode, showed significant contribution to EEG classification.
• The time window of 300-500 ms, peaking around 300 ms, was most critical for EEG classification.
• Joint attention training enhanced gradient contribution at 300 ms specifically in social scenarios.
• Increased training led to a forward shift in P300 latency in social, but not non-social, scenarios for ASD patients.

Conclusions:

• Joint attention training can effectively improve cognitive abilities related to social characteristics in individuals with ASD.
• EEGNet and saliency mapping provide valuable insights into the neural mechanisms underlying attentional improvements in ASD.
• The findings highlight the potential of targeted interventions to enhance social responsiveness in autism spectrum disorder.