Optimal selection of SOP and SPH using fuzzy inference system for on-line epileptic seizure prediction based on EEG phase synchronization

Living conditions of patients with refractory epilepsy will be significantly improved by a successful prediction of epileptic seizures. A proper warning impending seizure system should be resulted not only in high accuracy and low false positive alarms but also in suitable prediction time. In this study, the mean phase coherence index was used as a reliable indicator for identifying the pre-ictal period of 21-patient Freiburg dataset. In order to predict the seizures on-line, an adaptive Neuro-fuzzy model named ENFM (evolving Neuro-fuzzy model) was used to classify the extracted features. The ENFM was trained by a new class labeling method based on the temporal properties of a prediction characterized by two time intervals, seizure prediction horizon (SPH) and seizure occurrence period (SOP), which are subsequently applied in evaluation method. It is evident that increasing the SPH duration can be more beneficial to patients in preventing irreparable consequences of the seizure, as well as providing adequate time to deal with the seizure. In addition, a reduction in SOP duration can reduce the patient's stress in SOP interval. These two theories motivated us to design Mamdani fuzzy inference system considering sensitivity and FPR of the prediction result in order to find optimal SOP and SPH for each patient. 10-patient dataset assigned for optimizing the fuzzy system, while the rest of data was used to test the model. The results showed that mean SOP by 6 min and mean SPH by 27 min provided the best outcome, so that last seizure as well as about 15-h inter-ictal period of each patient were predicted on-line without false negative alarms, yielding on average 100% sensitivity, 0.13 per hour FPR, 86.95% precision and 92.5% accuracy.