Can EEG processing reveal seizure prediction patterns?

Epilepsy is characterized by an unexpected and frequent malfunction of the brain. Electrical activity in the brain has been studied for years in an attempt to predict seizures. This paper processes raw intracranial EEG recordings from different subjects in the time prior to seizure. A set of indicators is extracted from non-overlapping scrolling windows of I sec duration. The objective was to identify patterns that reveal that a seizure is developing before it occurs. While the exhaustive analysis did not detect patterns appropriate to predict a seizure, some indicators were observed to behave in time more similar independent of the subject. Similar time evolution was found for the activity and the power of the alpha and delta bands. It is also shown that the behavior of the correlation integral is somehow similar minutes before the seizure.