EEG-Based Mental Health Diagnosis: A Novel Approach Using DWT and Machine Learning

Mental health disorder impact and individual's cognitive and emotional well-being significantly. For making treatment to be effective, an early and accurate diagnosis is essential. This study presents an EEG-based mental health disorder classification system that extracts and analyzes brainwave features to identify potential mental health disorders. This system processes EEG signals using DWT (Discrete Wavelet Transform) to extract frequency band feature (Delta, Theta, Alpha, Beta and Gamma). These features are then classified using Deep learning models like Artificial Neural Networks (ANN), Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and hybrid CNN-LSTM and CNN-BiLSTM architectures. In this study raw data is first converted into frequency band and then dataset is preprocessed using Minmax Scaling, KNN imputation and feature label encoding and balance the dataset using random oversampling to ensure high data quality. Model performance is evaluated using validation accuracy, with hybrid CNN-BiLSTM achieving the highest accuracy. The proposed approach offers a fast and automated way to classify mental disorders using EEG data, making it a valuable tool for diagnosis and clinical decision-making. By analyzing brainwave patterns with machine learning, this system can assist healthcare professionals in identifying disorders more efficiently. In future we will aim to work on expanding real-time EEG processing capabilities and deploying a system for clinical use.