Rest and Active Arm Movement Classification Based on Subthalamic Nucleus LFPs for Parkinson's Disease Subtypes

Adaptive deep brain stimulation (DBS) is
expected to improve Parkinson’s disease (PD) symptoms. This
advance in neurostimulation strategy requires to interact with
pathological electrical signaling from basal ganglia circuitry
and to use this continuous feedback for controlling therapeutic
DBS. Further knowledge about how such signals change in
terms of clinical phenotypes and rest/movement states is
necessary to optimize this adaptive method and to bridge the
clinical needs with daily activities. This work assessed the
differences in local field potentials (LFP) spectrum recorded
from the subthalamic nucleus (STN) in rest and active arm
movement and classified these conditions for the different PD
clinical subtypes seeking to estimate the cost underlying the
early movement detection. Our results suggest there is a
tradeoff between the signal length and accuracy, in which the
subtypes play an important role in this performance. These
findings may define a crucial design variable for real-time
applications and allow future perspectives for tailored DBS
design.