Automated Depression Detection Using EEG Signals with Deep Learning and Attention Mechanism

Abstract

Depression, have emerged as significant global health concerns, affecting millions of people worldwide. Conventional schemes for diagnosing assessments of depression are primarily subjective in nature, which may lack precision and be prone to bias. As a response to these challenges, this study depicts a novel depression detection mechanism that leverages electroencephalography (EEG) signals and utilizes a sophisticated deep learning architecture. EEG signals, being non-invasive and capable of capturing real-time brain activity, offer a promising avenue for the objective assessment of mental health conditions like depression. However, due to the complex nature of EEG data, accurately identifying depressive patterns requires advanced processing and analytical methodologies. In this study, we developed a deep learning (DL) model consisting of stacked Long Short-Term Memory (LSTM) layers, stacked Gated Recurrent Units (GRU), and an Attention mechanism to detect depression effectively. The choice of architecture is motivated by the need to capture temporal dependencies in EEG signals, which are critical for recognizing subtle changes associated with depressive states. LSTM and GRU layers, known for their ability to handle long-term dependencies, form the backbone of our model, enabling the effective learning of relevant temporal patterns within the EEG data. Furthermore, the Attention mechanism enhances the model’s ability to focus on critical segments of the EEG sequences that exhibit depression-related characteristics, improving interpretability and robustness. We trained and evaluated our model on a dataset comprising EEG recordings from both healthy individuals and those diagnosed with clinical depression. The recommended scheme attained an impressive accuracy of 99.5%, significantly surpassing existing approaches in the field. This high accuracy underscores the model’s potential as a reliable tool for clinical applications, offering a promising step toward the automated, objective assessment of depression through EEG analysis.