Parkinsonian rest tremor can be distinguished from voluntary hand movements based on subthalamic and cortical activity using machine learning

Abstract

Tremor is one of the cardinal symptoms of Parkinsons disease. The neurophysiology of tremor is not completely understood, and so far it has not been possible to distinguish tremor from voluntary hand movements based on local brain signals. Here, we re-analyzed magnetoencephalography and local field potential recordings from the subthalamic nucleus of six patients with Parkinsons disease. Data were obtained after withdrawal from dopaminergic medication (Med Off) and after administration of levodopa (Med On). Using gradient-boosted tree learning, we classified epochs as tremor, self-paced fist-clenching, static forearm extension or tremor-free rest. While decoding performance was low when using subthalamic activity as the only feature (balanced accuracy mean: 38%, std: 7%), we could distinguish the four different motor states when considering cortical and subthalamic features (balanced accuracy mean: 75%, std: 17%). Adding a single cortical area improved classification by 17% on average, as compared to classification based on subthalamic activity alone. In most patients, the most informative cortical areas were sensorimotor cortical regions. Decoding performance was similar in Med On and Med Off. Our results demonstrate the advantage of monitoring cortical signals in addition to subthalamic activity for movement classification. By combining cortical recordings, subcortical recordings and machine learning, future adaptive systems might be able to detect tremor specifically and distinguish between several motor states.