Testing the null hypothesis of the nonexistence of a preseizure state

Abstract

A rapidly growing number of studies deals with the prediction of epileptic seizures. For this purpose, various techniques derived from linear and nonlinear time series analysis have been applied to the electroencephalogram of epilepsy patients. In none of these works, however, the performance of the seizure prediction statistics is tested against a null hypothesis, an otherwise ubiquitous concept in science. In consequence, the evaluation of the reported performance values is problematic. Here, we propose the technique of seizure time surrogates based on a Monte Carlo simulation to remedy this deficit.

Statistical validation of event predictors: A comparative study based on the field of seizure prediction

“…In the field of seizure prediction, it has been proposed to randomize either the seizure onset times [7], the alarms triggered by a predictor [5], or the time series of the features extracted from the EEG [8]. We emphasize that for other fields of event predictions, the same strategies can be followed when substituting seizures by events.…”

“…In seizure prediction, both analytical and Monte Carlo based validation approaches have been introduced [5][6][7][8][9][10][11][12][13].…”

“…In the following, we concentrate on the other methods. For so-called "seizure times surrogate" (STS), the time intervals between seizures are randomly permuted by drawing without replacement to generate randomized seizure onset times [7,10]. A random offset is added to the first interval because otherwise all realizations would add up to the same total duration and the last seizure onset times would always coincide with the original ones [10].…”

A signal regularity-based automated seizure prediction algorithm using long-term scalp EEG recordings

“…Several methods have been proposed for validating the performance of a seizure warning algorithm [37][38][39]. Each study tested a specific statistical hypothesis.…”

“…Nevertheless, there are several studies that have applied different random warning schemes or surrogate data to test different hypotheses [22,[37][38][39]. The comparison of the proposed algorithm with other random warning schemes should be followed.…”