Brain State Functional Network Analysis in Early Parkinson Disease Inferred by Hidden Markov Model

Parkinson's disease (PD) is frequently accompanied by motor symptoms such as tremor and bradykinesia, as well as non-motor symptoms, including cognitive impairment. It has been indicated that the reduction of synchronization level between different brain regions may lead to the occurrence of these symptoms. The degree of damage to the brain network may serve as a potential biomarker for describing the brain function of Parkinson's disease. Therefore, a comprehensive and individual analysis of the potential brain network mechanisms in earl Parkinson disease (ePD) is great of importance. Currently, exploring brain network changes relies on the sliding-window analysis, but brain networks established from the time series within a too long window may result in sub-second properties to be covered or even lost. In this work, we propose an innovative method to construct a hidden brain state functional network using a Hidden Markov Model (HMM). By dividing the window of the original Electroencephalogram (EEG) time series via estimated HMM series, sub-second individual information of the brain network state can be explored. We analyze and compare the connectivity, fluctuation and graph theory parameters of the hidden brain state functional network between the healthy control group and the ePD group, as well as their correlation with the clinical MoCA scale. It is found that in some brain states, functional connectivity disorders in the frontal lobe, occipital lobe and parietal lobe, difficulties in maintaining stable connectivity modes and the reduction of brain functional integration ability exist in PD group. Meanwhile, it is found that the degree of these situations will deepen with the decline of cognitive level.