Enhancing Diagnosis Precision in Alcohol Addiction Detection Through CNN Analysis with SMOTE-ENN Data Augmentation

Alcoholism is a major concern in modern culture, demanding effective and prompt detection measures to limit its negative consequences. Electroencephalography (EEG) data processing has emerged as a viable option for detecting alcoholism by utilizing the brain's unique electrical patterns. However, EEG signals are complicated and multi-channel, deciphering them is difficult. Previous research has primarily relied on traditional machine learning and statistical methodologies, frequently including handcrafted features to identify subjects as alcoholic or non-alcoholic. Deep learning models have gained enthusiasm as computer capabilities and data volumes have increased, suggesting prospective options to improve alcoholism diagnosis precision. This study looks into the efficacy of machine learning algorithms like Naive Bayes, Random Forest, and KNN, as well as deep learning algorithms like CNN and LSTM, when combined with data augmentation techniques to improve the performance of alcohol addiction detection. The research uses two independent datasets - one publicly available and one acquired - and performs rigorous preprocessing approaches, including the elimination of NaN values and duplicates, followed by normalization. Three data augmentation strategies, SMOTE, SMOTE-ENN, and SMOTE-TOMEK, are used to address class imbalances in the datasets. The results of a comprehensive research show significant advances in alcohol addiction detection. In the public dataset, the Random Forest method obtains 96.22% accuracy before augmentation, which rises to 98.67% after augmentation with SMOTE-ENN. Similarly, in the obtained dataset, the CNN model obtains 96% accuracy before augmentation, which increases to 98.97% after augmentation with SMOTE-ENN. These findings highlight the potential of deep learning models, when combined with good data augmentation methodologies, to greatly improve the accuracy and reliability of alcohol addiction diagnosis, providing vital insights for future research and clinical applications.