[Prediction of epilepsy based on common spatial model algorithm and support vector machine double classification]

Area of Science:

• Neurology
• Biomedical Engineering
• Signal Processing

Background:

• Epilepsy patient prediction is time-consuming and subjective.
• Existing electroencephalogram (EEG) analysis methods often lack accuracy.

Purpose of the Study:

• To develop an automatic and accurate epilepsy recognition method for EEG signals.
• To improve upon existing methods by incorporating spatial and time-frequency characteristics.

Main Methods:

• Utilized Common Spatial Pattern (CSP) for spatial feature extraction from EEG signals.
• Integrated standard deviation, entropy, and wavelet packet energy for comprehensive feature extraction.
• Implemented a novel double classification model based on Support Vector Machine (SVM).

Main Results:

• Achieved an average recognition rate of 98.73% for the first classification category.
• Attained an average recognition rate of 99.90% for the second classification category.
• Demonstrated significant improvement in distinguishing between interictal and ictal periods.

Conclusions:

• The proposed method effectively addresses limitations of previous epilepsy detection techniques.
• The combination of spatial features and a double SVM classification model enhances identification efficiency.
• This approach offers a reliable tool for epilepsy prediction and diagnosis.