Toward closed-loop optimization of deep brain stimulation for Parkinson's disease: concepts and lessons from a computational model

Abstract

Deep brain stimulation (DBS) of the subthalamic nucleus with periodic, high-frequency pulse trains is an increasingly standard therapy for advanced Parkinson's disease. Here, we propose that a closed-loop global optimization algorithm may identify novel DBS waveforms that could be more effective than their high-frequency counterparts. We use results from a computational model of the Parkinsonian basal ganglia to illustrate general issues relevant to eventual clinical or experimental tests of such an algorithm. Specifically, while the relationship between DBS characteristics and performance is highly complex, global search methods appear able to identify novel and effective waveforms with convergence rates that are acceptably fast to merit further investigation in laboratory or clinical settings.

Robust stabilization of delayed neural fields with partial measurement and actuation

“…As reported in the survey paper (Carron et al, 2013), several attempts have been made in that direction. These include adaptive and on-demand stimulation (Rosin et al, 2011;Graupe et al, 2010;Santaniello et al, 2011;Little et al, 2016;Marceglia et al, 2007), delayed and multi-site stimulation (Lysyansky et al, 2011;Batista et al, 2010;Pfister and Tass, 2010;Tass et al, 2012), optimal control strategies (Feng et al, 2007), and activity regulation (Haidar et al, 2016;Wagenaar et al, 2005;Grant and Lowery, 2013).…”

Mechanisms of Deep Brain Stimulation in Movement Disorders as Revealed by Changes in Stimulus Frequency

“…These assertions are supported by a recent computational study wherein methods were outlined to identify stimulus patterns that effectively override pathological activity in PD, but at lower average frequencies. 88 …”

Optimizing microstimulation using a reinforcement learning framework

“…Explicitly modeling and inverting a model of the neural response to microstimulation of varying parameters is alluring if the model can be adapted online [3]; however, a general model that encompasses all the parameters of spatio-temporal stimulation may be ill-posed without sufficient training data. If the possible stimulation space has many dimensions it is inefficient to naively sample the space to generate training data, and without sufficient data the model would fail to generalize the complex relationships between neural response and stimulation parameters [4]: making a modelfree approach appealing [5].…”

“…The amplitude is selected given the ongoing local potential at another predetermined electrode position to increase the reliable control of the evoked potentials. The authors in [5] use genetic algorithms to optimize the temporal waveform for deep-brain stimulation on a neural simulator. With microelectrode arrays, there is the possibility to modify both the spatial and temporal parameters of the stimulation.…”