Abnormality Analysis and UKF-based Estimation of Characteristic Features in Alzheimer's Disease: A Study Using a Neural Mass Model

Abstract

In this paper, thalamo-cortico-thalamic neural mass model(NMM) which is consist of excitatory thalamo-cortical Relay (TCR) and inhibitory interneurons (IN) neural populations is applied to mimic the changes of electroencephalograph (EEG) in Alzheimer's disease (AD). Spectrum analysis is proposed to study the effect of changes in synaptic connectivity on the power spectrum density (PSD) as well as average dominant frequency (DF) of the model output. It is observed that he synaptic connectivity parameters of the model play a crucial role in affecting the oscillatory behavior of the model output, and the PSD and average DF is increased with the excitatory parameter increased, while opposite result could be presented when the inhibitory parameter increased. Then unscented Kalman filter (UKF) method is applied to estimate the unobservable characteristic parameter of the NMM accurately and rapidly, with small relative error and dramatic convergent rate. The results presented in this paper may facilitate our understanding of the neural mechanisms underlying the alteration of EEG band power in AD brain, and become foundation of theoretical research of AD.