TFS-FENet: A Time-Frequency Spatial Deep Learning Framework for EEG-Based ADHD Subtype Classification

Researchers developed a deep learning model (TFS-FENet) that analyzes EEG signals to classify ADHD subtypes and distinguish ADHD from typical development with high accuracy (96.89% for three-class and 99.36% for binary classification). The approach uses convolutional neural networks to simultaneously capture time-frequency and spatial features of brain activity, positioning EEG-based machine learning as a potential objective biomarker to assist clinical diagnosis.