A novel fuzzy deep learning network for electroencephalogram classification of major depressive disorder.

Abstract

This study introduces the EEG-FDL model, a novel optimized fuzzy deep learning approach for classifying Major Depressive Disorder (MDD) using EEG data. Integrating deep learning with fuzzy learning via the Non-Dominated Sorting Genetic Algorithm II (NSGA-II), EEG-FDL optimizes fuzzy membership functions and backpropagation. The model handles noise and data uncertainty, achieving a remarkable 99.72% accuracy in distinguishing MDD from healthy EEG signals using 5-fold cross-validation on a large dataset. External validation further confirms its efficacy. EEG-FDL outperforms traditional classifiers due to its effective handling of uncertainties and optimized parameter tuning.