Parkinson's disease (PD) is a neurodegenerative disorder associated with alterations of neural activity and information processing primarily in the basal ganglia and cerebral cortex. Deep brain stimulation (DBS) of the subthalamic nucleus (STN-DBS) is the most effective therapy when patients experience levodopa-induced motor complications. A growing body of evidence points towards a cortical effect of STN-DBS, restoring key electrophysiological markers, such as excessive beta band oscillations, commonly observed in PD. However, the mechanisms of STN-DBS remain elusive. Here, we aim to better characterize the cortical substrates underlying STN-DBS-induced improvement in motor symptoms. We recorded electroencephalograms (EEG) from PD patients and found that, although apparent EEG features were not different with or without therapy, EEG signals could more accurately predict limb movements under STN-DBS. To understand the origins of this enhanced information transmission under STN-DBS in the human EEG data, we investigated the information capacity and dynamics of a variety of computational models of cortical networks. The extent of improvement in decoding accuracy of complex naturalistic inputs under STN-DBS depended on the synaptic parameters of the network as well as its excitability and synchronization levels. Additionally, decoding accuracy could be optimized by adjusting STN-DBS parameters. Altogether, this work draws a comprehensive link between known alterations in cortical activity and the degradation of information processing capacity, as well as its restoration under DBS. These results also offer new perspectives for optimizing STN-DBS parameters based on clinically accessible measures of cortical information processing capacity.
Keywords: deep brain stimulation; excitability; information processing; model; synchronization.