Efficient novel network and index for alcoholism detection from EEGs.

Area of Science:

• Neuroscience
• Biomedical Engineering
• Machine Learning

Background:

• Alcoholism is a severe condition causing significant brain damage and neurological, social, and behavioral issues.
• Current assessment methods like the Cut down, Annoyed, Guilty, and Eye-opener (CAGE) technique are time-consuming and prone to errors.

Purpose of the Study:

• To develop an automated system for alcoholism identification using electroencephalography (EEG) signals.
• To overcome the limitations of existing diagnostic methods.
• To aid practitioners and researchers in diagnosing alcoholism.

Main Methods:

• Investigated the Fast Walsh-Hadamard transform of EEG signals to analyze signal variability.
• Identified 36 linear and nonlinear features from EEG signals for pattern deciphering.
• Employed feature selection strategies and evaluated 19 machine learning algorithms and 5 neural network classifiers.

Main Results:

• Achieved 97.5% accuracy, 96.7% sensitivity, and 98.3% specificity using correlation-based feature selection with Recurrent Neural Networks.
• Attained 93.3% classification accuracy with novel matrix determinant features.
• Developed a unique index with clinically meaningful features for 100% accurate classification between healthy and alcoholic individuals.

Conclusions:

• The proposed automated system effectively identifies alcoholism using EEG signals.
• The novel index provides a highly accurate and efficient tool for real-time alcoholism detection.
• This technology can assist healthcare professionals and engineers in developing advanced diagnostic systems.