ANALYSIS OF MIGRAINE AURA BASED ON EEG TESTS USING ARTIFICIAL INTELLIGENCE METHODS

The dynamics of the membrane potential and recovery mechanism of biological neurons during migraine attacks with aura were mathematically modeled using the FitzHugh-Nagumo nonlinear stochastic differential equation system.The degree of influence of the membrane potential on the recovery mechanism, the degree of self-regulation of the recovery mechanism, and the stochastic resonance intensity coefficients affecting both components were determined using a fully connected neural network.This study addresses a significant gap in computational neuroscience by integrating stochastic differential equations with machine learning to characterize neuronal behavior during pathological conditions.Traditional electroencephalography (EEG) analysis methods primarily rely on time-frequency decomposition and statistical techniques, identifying fundamental signal characteristics but remaining disconnected from mechanistic stochastic neuronal models.Our approach combines the biophysically realistic FitzHugh-Nagumo framework with neural network-based parameter estimation using EEG recordings, enabling precise quantification of key biophysical parameters governing neuronal excitability and recovery processes.The identified parameters provide quantitative measures of membrane dynamics and stochastic fluctuations characterizing migraine pathophysiology, offering potential biomarkers for clinical diagnosis and personalized treatment strategies.