What is it about?

A method is proposed to predict epileptic seizures within a 15-minute window before the seizure onset. The prediction takes place by using electrical and optical signals extracted from the brain. The signals are classified with neural networks. The neural network is trained to identify the pre-ictal phase, which is the signal segment that occurs before the actual seizure.

Featured Image

Why is it important?

The application of this work is health care. The implementation of a portable device for seizure prediction will be useful to epileptic patients since this application enhances the quality of life of these patients. Research, related to this problem, is important because it also assists in the understanding of this problem.


My research efforts to solve the challenge of epileptic seizure prediction were wonderful experiences. Besides the satisfaction of contributing to the fields of Medicina, Neurology, Biomedical Signal Processing, Pattern Recognition, this research might be very easily be extended to solve other interesting and important problems such as stock market prediction (finance), Parkinson disease detection (health care), exo-planet detection and identification (astronomy).

Roberto Rosas-Romero
Universidad de las Américas-Puebla

Read the Original

This page is a summary of: Prediction of epileptic seizures using fNIRS and machine learning, Journal of Intelligent & Fuzzy Systems, February 2020, IOS Press,
DOI: 10.3233/jifs-190738.
You can read the full text:




The following have contributed to this page