Comprehensive Summary
This article introduces an advanced multimodal approach to improve emotion recognition in individuals with autism spectrum disorder (ASD). Currently, the traditional diagnostic and monitoring methods for autism only include behavioral observation which is very subjective and limited. Since emotion dysregulation is a central characteristic of ASD, the researchers designed an ensemble-based classification framework that includes electroencephalogram (EEG) and galvanic skin response (GSR) to analyze emotional responses quantitatively and mitigate subjective diagnoses. Autistic and neurotypical participants were exposed to a set parameter of emotional stimuli during which the EEG captured cortical neural activity and GSR measured autonomic arousal. The features included time-domain statistics, power spectral densities from distinct EEG frequency bands like delta, theta, alpha, beta, and gamma, and electrodermal conductivity changes reflecting sympathetic activation. This data was organized into different models like Support Vector Machine (SVM), Random Forest (RF), K-Nearest Neighbor (KNN), and Multilayer Perceptron (MLP) to minimize bias. The results revealed that combining EEG and GSR features substantially enhanced classification accuracy and emotional differentiation compared to using them alone. Specifically, the ensemble classifier achieved an accuracy of 98.5%, which outperformed the individual EEG-based (91.4%) and GSR-based (87.6%) models. It demonstrates strong classification reliability across multiple emotional states, specifically when distinguishing anxiety and neutral affect. Therefore the effects of the combination of EEG and GSR responses help reduce bias, noise sensitivity, and improve differentiation in emotion in autistic patients.
Outcomes and Implications
This has significant medical implications because it provides early detection, diagnosis, and therapeutic monitoring of autism spectrum disorder. The study provides a quantitative and more objective biomarker to emotion that can help with diagnosing along with the DSM-5 criteria. The 98.5% accuracy allows for future steps in real-world integration by providing wearable devices or portable monitoring systems that can track emotion, according to the researchers. The EEG-GSR fusion provides physicians with quantifiable insights into what types of stimuli affect an autistic brain and how they can monitor emotional fluctuation in children with ASD. The researchers hope to get larger datasets and longitudinal validation to ultimately perform remote diagnostics and visualize a patient’s physiological responses in real time and adjust therapy intensity accordingly. Overally, the EEG-GSR combination network has the ability to capture both the brain and autonomic signatures of emotion to provide earlier, more individualized, and less invasive autism care in the future.