How well can epileptic seizures be predicted? An evaluation of a nonlinear method.

The unpredictability of the occurrence of epileptic seizures contributes to the burden of the disease to a major degree. Thus, various methods have been proposed to predict the onset of seizures based on EEG recordings. A nonlinear feature motivated by the correlation dimension is a seemingly promising approach. In a previous study this method was reported to identify 'preictal dimension drops' up to 19 min before seizure onset, exceeding the variability of interictal data sets of 30-50 min duration. Here we have investigated the sensitivity and specificity of this method based on invasive long-term recordings from 21 patients with medically intractable partial epilepsies, who underwent invasive pre-surgical monitoring. The evaluation of interictal 24-h recordings comprising the sleep-wake cycle showed that only one out of 88 seizures was preceded by a significant preictal dimension drop. In a second analysis, the relation between dimension drops within time windows of up to 50 min before seizure onset and interictal periods was investigated. For false-prediction rates below 0.1/h, the sensitivity ranged from 8.3 to 38.3% depending on the prediction window length. Overall, the mean length and amplitude of dimension drops showed no significant differences between interictal and preictal data sets.

The seizure prediction characteristic: a general framework to assess and compare seizure prediction methods.

The unpredictability of seizures is a central problem for all patients suffering from uncontrolled epilepsy. Recently, numerous methods have been suggested that claim to predict from the EEG the onset of epileptic seizures. In parallel, new therapeutic devices are in development that could control upcoming seizures provided that their onset is known in advance. A reliable clinical application controlling seizures, consisting of a seizure prediction method and an intervention system, would improve patient quality of life. The question therefore arises as to whether the performance of the seizure prediction methods is already sufficient for clinical applications. The answer requires assessment criteria to judge and compare these methods, but recognized criteria still do not exist. Based on clinical, behavioral, and statistical considerations, we suggest the "seizure prediction characteristic" to evaluate seizure prediction methods. Results of this approach are exemplified by its application to the "dynamical similarity index" seizure prediction method using 582 hours of intracranial EEG data, including 88 seizures.

Reset of human neocortical oscillations during a working memory task.

Both amplitude and phase of rhythmic slow-wave electroencephalographic activity are physiological correlates of learning and memory in rodents. In humans, oscillatory amplitude has been shown to correlate with memory; however, the role of oscillatory phase in human memory is unknown. We recorded intracranial electroencephalogram from human cortical and hippocampal areas while subjects performed a short-term recognition memory task. On each trial, a series of four list items was presented followed by a memory probe. We found agreement across trials of the phase of oscillations in the 7- to 16-Hz range after randomly timed stimulus events, evidence that these events either caused a phase shift in the underlying oscillation or initiated a new oscillation. Phase locking in this frequency range was not generally associated with increased poststimulus power, suggesting that stimulus events reset the phase of ongoing oscillations. Different stimulus classes selectively modulated this phase reset effect, with topographically distinct sets of recording sites exhibiting preferential reset to either probe items or to list items. These findings implicate the reset of brain oscillations in human working memory.