English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Open

Presents the pdf analysis as premium feature

PDF Analysis

Insights

Presents the summarisation as premium feature

Summarisation

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

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 electroencephalo- gram 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. Epilepsies are characterized by recurrent and often severe malfunctions of the brain that manifest themselves as epilep- tic seizures. Most epilepsy patients experience the onset of a seizure as a sudden and unexpected event. Guided by both a priori and a posteriori considerations, however, it has been hypothesized that the transition to the seizure ~ictal! state might not be an abrupt phenomenon but rather evolves via a temporally extended preictal state ~e.g., Ref. @1#!. Provided that such a preictal state detection could be achieved with a sufficient sensitivity and specificity, seizure anticipation and prevention technologies could be envisaged which would be of great benefit for epilepsy patients. In Refs. @2-4 # ,i t has been investigated whether information about an impending seizure can be extracted from the electroencephalogram ~EEG! using different characterizing measuresderived from linear or nonlinear time series analysis. Common to these studies is a two-step procedure: First, a characterizing mea- sure is calculated for a multichannel EEG using a moving- window technique. In a second step, the resulting spatiotem- poral profile of the characterizing measure is analyzed by means of an often highly elaborated evaluation scheme aim- ing at an extraction of information specific for the preictal state. As distinct and complementary as the different ap- proaches are, in the context of the present study, they will be termed as seizure prediction statistics in their collectivity. Their output in terms of sensitivity and specificity will be denoted as performance. Let us now consider the following null hypothesis: ''The transition from the interictal to the ictal state is an abrupt phenomenon. An intermediate preictal state does not exist.'' Despite the fact that in this case, no information predictive of impending seizures could be extracted from the EEG, many of the seizure prediction statistics would probably still render nonzero performance values. Moreover, an a priori estima- tion of these performance values is problematic. Hence, it is impossible to decide whether a given performance value ob- tained from real data indicates the existence of a preictal state or whether it is consistent with the null hypothesis stated above. A similar problem is known from the application of non- linear time series analysis techniques to stochastic dynamics. The framework of nonlinear time series analysis comprises of a number of measures that allow the characterization of nonlinear deterministic dynamics @5#. For most of these mea- sures, however, the range of values obtained for nonlinear deterministic dynamics and for linear stochastic dynamics overlap substantially @6#. It is, therefore, impossible to decide

Testing the null hypothesis of the nonexistence of a preseizure state