Publication | Open Access

Removing artefacts and periodically retraining improve performance of neural network-based seizure prediction models