An automated drug dependence detection system based on EEG

Area of Science:

• Neuroscience
• Medical Technology
• Data Science

Background:

• Substance abuse significantly impacts brain structure and function.
• Developing objective diagnostic tools for substance use disorders is crucial.

Purpose of the Study:

• To design an automated system for detecting drug dependence in multidrug (MD) abusers using electroencephalogram (EEG) signals.
• To investigate the alterations in EEG signal complexity associated with MD abuse.

Main Methods:

• Recorded EEG signals from 10 MD-dependent individuals and 12 healthy controls (HC).
• Analyzed EEG dynamic characteristics using Recurrence Plot and quantified complexity via entropy index (ENTR).
• Employed Support Vector Machine (SVM) for classification of MD abusers and HC groups.

Main Results:

• MD abusers exhibited decreased ENTR in delta, alpha, beta, gamma, and all-band EEG signals, indicating reduced complexity.
• An increased ENTR in the theta band was observed in MD abusers.
• The SVM classifier achieved 90% accuracy, 89.36% sensitivity, 90.7% specificity, and 89.8% F1 score in distinguishing the groups.

Conclusions:

• Nonlinear analysis of EEG signals can aid in the automatic diagnosis of MD abuse.
• The developed system demonstrates potential as a diagnostic aid for identifying individuals with MD dependence.