AI Summary & Key Takeaways
AI InsightThe results indicate that the SVM algorithm achieved an accuracy of 98.16% in individual channels.
📌 3 Key Takeaways
Abstract
This study proposes an innovative approach to analyze EEG (electroencephalography) signals for the diagnosis of paranoid schizophrenia. It introduces a novel feature extraction method called the Ompat pattern, which represents EEG signals as a directed graph within a 6×6-dimensional structure to capture specific signal patterns. The method extracts 20-bit features from each edge, generating a total of 20 edges that provide meaningful data for classification. K-nearest neighbors (KNN), Support Vector Machine (SVM), Random Forest, and Fine Tree algorithms are utilized in the classification phase. The classification performance across different EEG channels is evaluated, and accuracy rates are optimized through the majority voting method. The results indicate that the SVM algorithm achieved an accuracy of 98.16% in individual channels. When majority voting is applied, the accuracy increases to 98.25%, with the best performance observed in channel 12. The experimental results suggest that the proposed method has the potential to provide an automated and objective approach for diagnosing paranoid schizophrenia, thereby supporting clinical diagnostic processes.
Özet
This study proposes an innovative approach to analyze EEG (electroencephalography) signals for the diagnosis of paranoid schizophrenia. It introduces a novel feature extraction method called the Ompat pattern, which represents EEG signals as a directed graph within a 6×6-dimensional structure to capture specific signal patterns. The method extracts 20-bit features from each edge, generating a total of 20 edges that provide meaningful data for classification. K-nearest neighbors (KNN), Support Vector Machine (SVM), Random Forest, and Fine Tree algorithms are utilized in the classification phase. The classification performance across different EEG channels is evaluated, and accuracy rates are optimized through the majority voting method. The results indicate that the SVM algorithm achieved an accuracy of 98.16% in individual channels. When majority voting is applied, the accuracy increases to 98.25%, with the best performance observed in channel 12. The experimental results suggest that the proposed method has the potential to provide an automated and objective approach for diagnosing paranoid schizophrenia, thereby supporting clinical diagnostic processes.