Assessing Heart Rate Variability Changes Ahead of an Epileptic Seizure. Outcomes from a Multivariate Analysis.
AbstractPublished evidence suggests physiological changes before the onset of an epileptic seizure. However little is known as per when these changes start. This research was aimed at describing heart rhythm changes in the period from 57 minutes to just the onset of an epileptic seizure. Electrocardiograms corresponding to preictal condition before seven partial seizures were downloaded from the publicly available dataset at physionet.org. Five-min segments taken at different moments were submitted to heart rate variability (HRV) analysis using the Kubios 2.2 HRV package. Twenty five HRV variables were submitted to multivariate regression analysis with “Time_to_seizure_onset” as the variable to be predicted. Discriminant analysis was implemented to separate closer to the seizure segments from those appearing much earlier. Usually about eight HRV variables were picked into the optimal predictive subset. Correlations between 0.76 and 0.92 (CI for p>0.05) were obtained for individual seizures as well as for a pooled data set when observed times before seizure were compared to predictions from multivariate regression. Discriminant analysis separated nearer (0-25min before the seizure) from farther (26-57 min) HRV data with a precision of 79.7%. Our results suggest that cardiac rhythm changes start around one hour before the seizure, with underlying dynamics changing as the time to seizure onset is being approached. These results can have both theoretical and practical implications for seizure prediction and epilepsy treatment.