Review on Epileptic Seizure Prediction: Machine Learning and Deep Learning Approaches

Area of Science:

• Neurology
• Biomedical Engineering
• Computational Neuroscience

Background:

• Epileptic seizures, caused by brain abnormalities, occur abruptly, increasing mortality.
• Approximately 1% of the global population is affected by epilepsy.
• Predicting seizures aids in timely medication and prevention.

Purpose of the Study:

• To review automated schemes for detecting interictal spikes and epileptic seizures from EEG signals.
• To address challenges in feature selection and classification for epilepsy prediction algorithms.
• To provide insights into current techniques and future research directions in automated seizure prediction.

Main Methods:

• Review of various automated techniques for EEG signal analysis.
• Focus on machine learning and deep learning approaches.
• Examination of different feature selection and classification methods.

Main Results:

• EEG signals are often corrupted by noise and artifacts (e.g., eye blinks, muscle movements).
• Visual inspection of long EEG recordings is cumbersome and prone to errors.
• Automated detection of interictal spikes and seizures is crucial for precise analysis.

Conclusions:

• Feature selection and classification remain key challenges in epilepsy prediction.
• Automated schemes are essential tools for neurologists to categorize epileptic and non-epileptic signals.
• This review offers a comprehensive understanding of seizure prediction techniques and future research avenues.