Classification of Epileptic Seizure Using Hybrid Deep Learning Framework with Time and Time-Frequency Hjorth Features

This study proposes a machine learning framework combining Hjorth parameters extracted from time and time-frequency domains with deep neural networks (CNN, BiLSTM, and attention mechanisms) to classify epileptic seizure stages. The approach achieved 98.4% accuracy for binary classification and 85.4% for five-class discrimination on the Bonn EEG dataset using rigorous cross-validation.