Detection Analysis of Epileptic EEG Using a Novel Random Forest Model Combined With Grid Search Optimization

Area of Science:

• Neuroscience
• Biomedical Engineering
• Machine Learning

Background:

• Epileptic seizure detection is crucial for critically ill patients in intensive care units.
• Electroencephalogram (EEG) signal analysis is increasingly used for seizure detection.

Purpose of the Study:

• To develop and train a novel random forest algorithm for EEG decoding.
• To visualize EEG time-frequency features for improved seizure detection.
• To classify three different levels of epileptic conditions: healthy, seizure-free, and seizure activity.

Main Methods:

• Proposed an automatic detection framework using multiple time-frequency analysis approaches.
• Employed a novel random forest model with grid search optimization.
• Utilized short-time Fourier transformation for feature visualization and principal component analysis for dimensionality reduction.
• Classified 500 raw EEG samples with multiple cross-validations.

Main Results:

• The model achieved effective classification compared to previous methods.
• Evaluations included accuracy, confusion matrix, ROC curve, and AUC.
• Demonstrated improved detection performance and diagnostic accuracy.

Conclusions:

• The proposed computer-assisted diagnosis scheme shows potential guiding significance for clinical practice.
• This approach can help alleviate patient suffering and reduce neurologist workload.
• The framework offers a promising tool for objective epilepsy diagnosis and management.