Automatic identification of schizophrenia employing EEG records analyzed with deep learning algorithms

Area of Science:

• Neuroscience
• Medical Informatics
• Artificial Intelligence

Background:

• Electroencephalography (EEG) records brain's electrical activity for diagnosing mental disorders.
• Schizophrenia diagnosis can be challenging, necessitating advanced analytical tools.
• Current diagnostic methods may benefit from objective, data-driven approaches.

Purpose of the Study:

• To present a novel system for classifying schizophrenia patients using EEG recordings.
• To develop an algorithm that effectively distinguishes schizophrenia patients from healthy controls.
• To evaluate the performance of the proposed system against established machine learning algorithms.

Main Methods:

• Decomposition of EEG signals into radial basis functions via fuzzy means.
• Utilizing information from multiple EEG electrodes for classification.
• Comparison with K-Nearest Neighbor, Adaboost, Support Vector Machine, and Bayesian Linear Discriminant Analysis.

Main Results:

• The proposed fuzzy means decomposition method achieved high performance metrics.
• Balanced accuracy, recall, precision, and F1 score were all approximately 93%.
• The novel system demonstrated superior classification performance compared to classical algorithms.

Conclusions:

• The developed system offers a highly accurate method for classifying schizophrenia from EEG data.
• The algorithm's effectiveness in separating patients from controls is confirmed.
• This brain activity analysis model shows promise for aiding in schizophrenia prediction.