Deep Learning and Optimization Approaches in EEG-based Classification of Neuropsychiatric Disorders Using Deep Sparse Neural Networks with Gooseneck Barnacle Optimization Algorithm: A Review

Abstract

Electroencephalography (EEG)-based analysis has emerged as a powerful non-invasive tool for diagnosing neuropsychiatric disorders such as depression, schizophrenia, and bipolar disorder. Recent advances in deep learning (DL) have significantly improved the ability to extract complex temporal–spatial features from EEG signals, enabling more accurate and automated classification systems. This review explores deep learning and optimization approaches for EEG-based neuropsychiatric disorder classification, with a focus on deep sparse neural networks (DSNNs) and bio-inspired optimization techniques such as the Gooseneck Barnacle Optimization Algorithm (GBOA). Sparse neural architectures reduce computational complexity and enhance generalization, while optimization algorithms improve parameter tuning and convergence efficiency. The paper systematically reviews studies published between 2020 and 2023, highlighting trends in convolutional neural networks (CNNs), recurrent neural networks (RNNs), hybrid models, and graph-based architectures. Comparative analysis demonstrates that optimized DL models outperform traditional machine learning techniques in accuracy and robustness. However, challenges such as data variability, limited datasets, and interpretability remain. The integration of DSNNs with advanced optimization techniques presents a promising direction for scalable, efficient, and clinically applicable EEG-based diagnostic systems. This review provides insights into current advancements and identifies future research opportunities in neuropsychiatric disorder detection.