Automatic Recognition of High-Density Epileptic EEG Using Support Vector Machine and Gradient-Boosting Decision Tree
Jiaxiu He1, Li Yang1, Ding Liu1
1Department of Neurology, The Third Xiangya Hospital of CSU, Tongzipo Street, Changsha 410013, China.
Brain Sciences
|September 23, 2022
Summary
Gradient-boosting decision tree (GBDT) outperforms Support Vector Machine (SVM) in recognizing epileptic electroencephalogram (EEG) signals. GBDT achieved higher accuracy and F1-score, demonstrating its effectiveness in machine learning for epilepsy diagnosis.
Area of Science:
- Neurology
- Artificial Intelligence
- Biomedical Engineering
Background:
- Epilepsy (Ep) is a chronic neurological disorder diagnosed via seizure history and electroencephalogram (EEG).
- Machine learning (ML) models are increasingly applied for automated epileptic EEG recognition.
- High-density EEG data acquisition is crucial for accurate diagnostic analysis.
Purpose of the Study:
- To compare the classification performance of Support Vector Machine (SVM) and Gradient-Boosting Decision Tree (GBDT) algorithms.
- To evaluate classifier efficacy using controlled EEG data sources and feature sets.
- To identify the optimal machine learning model for epileptic EEG detection.
Main Methods:
- Utilized high-density EEG data collected from Xiangya Third Hospital.
- Extracted time-domain (EMD-processed), frequency-domain (PSD), and non-linear (Shannon entropy) features.
- Implemented and compared SVM and GBDT classifiers for epileptic EEG recognition.
Main Results:
- GBDT classifier achieved a superior accuracy of 90.00% and an F1-score of 93.40%.
- SVM classifier yielded an accuracy of 72.00% and an F1-score of 82.28%.
- GBDT demonstrated higher sensitivity (98.57%), precision (89.13%), and AUC (0.9119).
Conclusions:
- Gradient-Boosting Decision Tree (GBDT) is more effective than Support Vector Machine (SVM) for classifying epileptic EEG.
- Feature selection and classifier parameter control are key for optimizing ML model performance in epilepsy diagnosis.
- This study highlights GBDT's potential in advancing automated epileptic EEG analysis.
