A Patient-Specific Approach for Short-Term Epileptic Seizures Prediction Through the Analysis of EEG Synchronization

Area of Science:

• Neurology
• Biomedical Engineering
• Signal Processing

Background:

• Epilepsy affects over 65 million people globally, characterized by abnormal brain electrical activity.
• Accurate seizure prediction is crucial for developing effective monitoring and control devices.

Purpose of the Study:

• To develop a patient-specific, noninvasive approach for short-term epileptic seizure prediction.
• To explore the feasibility of using electroencephalogram (EEG) synchronization patterns for real-time seizure detection.

Main Methods:

• Utilized noninvasive EEG data to analyze brain synchronization patterns.
• Developed computationally inexpensive graph-based functions to quantify synchronization variations.
• Employed classifiers, including a novel threshold-based method, to distinguish preictal from interictal states.

Main Results:

• Demonstrated that the proposed method effectively highlights synchronization changes preceding seizures.
• Validated the approach on public EEG databases and patient data from the University of Siena.
• Showcased the efficacy of a simple, computationally viable processing technique.

Conclusions:

• The developed approach offers a promising step towards portable and wearable seizure prediction devices.
• Low computational requirements and noninvasive techniques make it suitable for real-life applications.
• Highlights the potential of analyzing EEG synchronization for epilepsy management.