PREDICTING THE COURSE OF EPILEPSY USING A SET OF PHYSIOLOGICAL PARAMETERS

Abstract

Aim. To predict the course of epilepsy using a set of physiological parameters.Materials and methods. We examined 72 healthy individuals (control group) and 163 patients with epilepsy by monitoring the EEG, the evoked potentials, and the parameters of the motor and vegetative systems. Based on the cluster approach, the patients were subdivided into subgroups according to their clinical and psychological characteristics, as well as the quality of life. These subgroups were quantitatively described using the logit regression models and the artificial neural networks technology.Results. We were able to discern between the subgroups of patients with a favorable and unfavorable course of the disease. The patients with an unfavorable course of epilepsy had an increased latency of cognitive evoked potentials, a decreased activation of the associative and the motor cerebral mechanisms, as well as a prevalence of the sympathetic activity. We have found reasonably good correlations between the patients’ clinical characteristics and the physiological parameters based on the logit regression analysis and artificial neural networks models.Conclusion. The clinical and psychosocial heterogeneity of patients with epilepsy is associated with the prevalence of symptomatic forms in the group with an unfavorable course of the disease. The crucial role in dividing the patients into clinical groups is played by the manifestations of nonspecific modulating brain structures.

Keywords

• epilepsy
• cluster analysis
• disease prognosis
• logit regression analysis
• artificial neural networks