Modelo no lineal basado en Wilson-Cowan y técnicas de simulación M/EEG para la predicción temprana de la progresión del Alzheimer usando datos ADNI

Abstract

Alzheimer’s disease is characterized by progressive alterations in structural and functional brain connectivity. This study proposes a personalized neurocomputational framework that integrates structural and diffusion magnetic resonance imaging to construct individual connectomes, which serve as the basis for implementing a population-based dynamic model grounded in the Wilson-Cowan system. From this model, M/EEG-like signals are simulated and spectral biomarkers, functional connectivity metrics, and dynamic parameters are extracted to characterize patterns associated with different clinical stages (CN, EMCI, MCI, and AD). The resulting features are subsequently integrated into a machine learning model aimed at clinical stage classification to assess the predictive potential of the proposed approach. The findings demonstrate the feasibility of constructing personalized brain models that integrate structural connectivity, neuronal dynamics, and electrophysiological simulation, establishing a reproducible methodology for the computational study of Alzheimer’s disease progression.