We can’t show the full text here under this license. Use the link below to read it at the source.
A scheme combining feature fusion and hybrid deep learning models for epileptic seizure detection and prediction
Combining feature mixing and hybrid deep learning to detect and predict epileptic seizures
AI simplified
Abstract
The proposed method achieved a of 99.24% and of 99.51% in seizure detection.
- A new method for seizure detection and prediction was developed using a convolutional neural network-gated recurrent unit-attention mechanism.
- (EEG) signals were processed through wavelet decomposition, resulting in six subbands for analysis.
- Features were extracted from the time-frequency domain and nonlinear characteristics of each subband.
- The method demonstrated an accuracy of 99.35% in detecting seizures and 95.16% in predicting them.
- Tenfold cross-validation on the CHB-MIT dataset validated the effectiveness of this approach.
AI simplified
Key numbers
99.24%
for Seizure Detection
Average achieved in seizure detection across 24 cases.
99.51%
for Seizure Detection
Average achieved in seizure detection across 24 cases.
95.47%
for Seizure Prediction
Average achieved in seizure prediction across 24 cases.