Epilepsy prediction study based on complex brain networks and Fiedler value analysis

Epilepsy, a chronic neurological disorder, profoundly affects patients’ quality of life and overall health. The accurate prediction of seizures is crucial for effective patient management and treatment strategies. Past research has shown that epileptic seizures are closely linked to abnormal, synchronized neuronal discharges within the brain. In this paper, we introduce an innovative seizure prediction approach that combines complex brain network analysis with Fiedler eigenvalue assessment. Initially, epileptic EEG data were preprocessed using advanced filtering techniques, and 23 bipolar leads were considered as individual nodes, which were subsequently categorized into six distinct brain regions. Subsequently, we calculated the transfer entropy between each electrode’s signals, followed by applying a thresholding process to generate an adjacency matrix, thus enabling the construction of the brain network and the computation of its Fiedler value. Standard curves depicting the temporal evolution of Fiedler values were obtained through polynomial fitting and compared with actual patient-specific fitted curves using dynamic time warping (DTW) distance calculations. Our findings indicate a significant similarity between most standard and actual curves, affirming the validity and precision of our proposed methodology.