TABS: Transformer Based Seizure Detection

In this paper, we develop and submit a novel model to the Neureka 2020 challenge
¹¹
https://neureka-challenge.com/ that can predict seizures from an EEG signal. This challenge places emphasis on limiting the number of false positives, or reducing the places in which background signal is erroneously labeled as a seizure. As a result, our model is intended to act as a tool to aid neurologists as opposed to supplanting them completely. The goal is to allow a doctor to move more quickly through EEG data, enabling the doctor to treat more patients and devote more time toward treatment. Our model is a transformer based neural network architecture. We call it TABS, after Transformer Based Seizure detector. TABS scored a sensitivity of 9.03% and a false alarm rate of 31.21 per 24 hours when evaluated using the Time-Aligned Event Score.