Objective: This study aimed to characterize dynamic brain states during freezing of gait in Parkinson’s disease using Hidden Markov Modelling of EEG data.

Background: Freezing of gait (FOG) is a disabling symptom of Parkinson’s disease (PD) characterized by sudden interruptions of locomotion. Increasing evidence suggests that FOG arises from abnormal interactions between large-scale brain networks (Asher et al., 2021). However, most neurophysiological studies rely on time-averaged measures and are limited in capturing the dynamic fluctuations of brain activity during FOG. Hidden Markov models (HMMs) provide a data-driven approach to identify transient brain states and quantify their temporal dynamics (Vidaurre et al., 2018), and have recently been applied to investigate brain-state dynamics in neurodegenerative disorders (Sharma et al., 2021).

Method: A total of 276 FOG trials from 13 PD patients were identified using 12-s epochs (-6s to +6s asround freezing onset). Data were preprocessed, source reconstructed (sLORETA) and parcellated using the Desikan-Killiany atlas. Concatenated source-level time series were modelled using a time-delay embedded Hidden Markov Model (TDE-HMM) was applied to identify transient brain states. State temporal properties (fractional occupancy, mean lifetime, and transition probabilities) were then quantified.

Results: The HMM identified six transient brain states during FOG with distinct temporal properties. State 1 showed the highest fractional occupancy and longest mean lifetime, indicating a dominant and stable network configuration. This state was characterized by widespread frontoparietal connectivity. In contrast, states 4-6 occurred less frequently and exhibited shorter lifetimes, representing brief transient network configurations with relatively stronger posterior cortical involvement. Transition probability revealed frequent returns to State 1, suggesting attractor-like dynamics. This pattern may reflect altered state switching during FOG, with brief transient states or a shift toward a dominant frontoparietal configuration consistent with increased cortical control of locomotion.

Conclusion: These findings suggest that FOG may involve pathological stabilization of cortical network dynamics, where the system becomes trapped in a dominant state rather than flexibly transitioning between configurations required for normal locomotion.

Figure 1

References: Asher, E. E., Plotnik, M., Günther, M., Moshel, S., Levy, O., Havlin, S., . . . Bartsch, R. P. (2021). Connectivity of EEG synchronization networks increases for Parkinson’s disease patients with freezing of gait. Communications biology, 4(1), 1017.
Sharma, A., Vidaurre, D., Vesper, J., Schnitzler, A., & Florin, E. (2021). Differential dopaminergic modulation of spontaneous cortico–subthalamic activity in Parkinson’s disease. elife, 10, e66057.
Vidaurre, D., Abeysuriya, R., Becker, R., Quinn, A. J., Alfaro-Almagro, F., Smith, S. M., & Woolrich, M. W. (2018). Discovering dynamic brain networks from big data in rest and task. Neuroimage, 180, 646-656.

« Back to 2026 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/dynamic-neural-states-underlying-freezing-of-gait-in-parkinsons-disease/