Electroencephalogram Based Depression Detection Using Ensemble Approach

Numerous people Shave lately been diagnosed with mental diseases such as depression all around the globe. The conventional methodology for distinguishing depression depends on the clinical questionnaire, which relies upon the patients’ reactions as well as detecting their behavior exercises. Recent advances in EEG and processing speed, along with well-trained ML models, have resulted in a potent method for diagnosing psychological disorders like depression, anxiety, and many more. Signals captured by the EEG equipment can measure potential changes between different areas of the cortex. EEG readings are a useful technique for diagnosing depression and are said to be more accurate than other signals. In this paper, we implemented the ensemble approach to explore depression recognition. In an experiment, 53 individuals took part, including 24 depressed patients and 29 healthy controls. We divided the extracted features from the EEG data into two: linear features and non-linear features. In this study, we developed an ensemble approach for detecting depression from EEG signals. We extracted linear and nonlinear features from the EEG data and used dimensionality reduction with LDA to improve classification performance. Our approach achieved a 91% accuracy rate in identifying MDD patients, outperforming previous studies. Our findings suggest that EEG-based machine learning is a promising new method for diagnosing depression. Future studies are needed to validate our approach in larger clinical trials.