Seizure Prediction Based on Transformer Using Scalp Electroencephalogram

Epilepsy is a chronic and recurrent brain dysfunction disease. An acute epileptic attack will interfere with patient’s normal behavior consciousness, having great impact on their life. The purpose of this study was to design seizure prediction model improve the quality patients’ lives assist doctors in making diagnostic decisions. This paper presents transformer-based model. Firstly, time-frequency characteristics electroencephalogram (EEG) signals were extracted by short-time Fourier transform (STFT). Secondly, three transformer tower used fuse classify features EEG signals. Finally, when combined attention mechanism networks, signal processed as whole, which solves problem length limitations deep learning models. Experiments conducted Children’s Hospital Boston Massachusetts Institute Technology database evaluate performance experimental results show that, compared previous classification models, our can enhance ability use time, frequency, channel information from accuracy prediction.