Enhanced hybrid deep neural network for EEG-based schizophrenia diagnosis using functional and temporal features

Area of Science:

• Neuroscience
• Psychiatry
• Artificial Intelligence

Background:

• Schizophrenia diagnosis relies on subjective clinical assessments, facing limitations in accuracy and efficiency.
• Electroencephalography (EEG) offers objective brain activity measures, showing potential for improved schizophrenia detection.
• Developing automated diagnostic tools is crucial to overcome traditional method drawbacks.

Purpose of the Study:

• To develop and validate a novel hybrid deep neural network for accurate schizophrenia diagnosis using EEG signals.
• To overcome the subjectivity and time constraints associated with traditional diagnostic methods.
• To enhance the efficiency and objectivity of schizophrenia detection.

Main Methods:

• Collected EEG data from individuals with schizophrenia and healthy controls during a visual task.
• Preprocessed EEG signals, extracted functional and time-domain features.
• Utilized a hybrid deep neural network, benchmarked against SVM and KNN, with k-fold cross-validation.

Main Results:

• The hybrid deep neural network demonstrated superior performance in distinguishing schizophrenia patients from controls.
• Achieved high diagnostic accuracy, sensitivity, and discriminative power.
• Outperformed established machine learning methods like support vector machines and k-nearest neighbors.

Conclusions:

• EEG-based diagnosis using a hybrid deep neural network is a promising approach for accurate schizophrenia detection.
• This method offers a more objective and efficient alternative to traditional clinical assessments.
• Further research should focus on dataset expansion and clinical validation for real-world application.