INFLUENCE OF AEROBIC EXERCISE AND DIETARY INTERVENTION ON EXECUTIVE FUNCTION IN OBESE CHILDREN: DEPTHWISE SEPARABLE CONVOLUTIONAL NETWORK-BASED ELECTROENCEPHALOGRAM ANALYSIS

Introduction: Childhood obesity not only impairs physical health but also adversely affects executive function. Although aerobic exercise and dietary interventions can improve cognition, traditional assessment methods are inadequate for capturing real-time brain activity. Moreover, there is a lack of research utilizing deep learning to analyze the electroencephalogram (EEG) for investigating the neural mechanisms underlying executive function in children with obesity. Methods: This study enrolled 100 students (50 in the obese group, 50 in the normal-weight group), all of whom underwent aerobic exercise training and dietary intervention. Executive function was assessed before and after the intervention, and EEG signals were collected. A deep separable convolutional network (DSCN) algorithm was employed to analyze the EEG data to improve classification accuracy and feature extraction efficiency. Results: the DSCN algorithm demonstrated significantly higher performance in EEG signal classification across accuracy, precision, recall, F1-score and AUC compared to traditional algorithms ([Formula: see text]). After the intervention, both behavioral indicators of executive function and EEG features showed significant improvement in the obese group, with effect sizes (Cohen’s [Formula: see text] ranging from medium to large (0.65–0.90). Discussion: The aerobic exercise training system integrated with the intelligent DSCN algorithm effectively enhances executive function in children with obesity. The DSCN-based EEG analysis tool offers a novel approach for studying the neural mechanisms of exercise interventions, demonstrating significant theoretical innovation and potential for clinical application.