Iterative feedback tuning of proportional-integral controller parameters for adaptive deep brain stimulation

Deep brain stimulation (DBS) is a well-established method for symptomatic treatment of Parkinson’s disease and essential tremor. Adaptive deep brain stimulation has the potential to surpass the performance of conventional DBS, providing more accurate symptom suppression, better control of stimulation-induced side effects, and longer battery life. While multiple controllers have been proposed and successfully tested in computational models as well as in patients, even the simple methods still require parameter tuning and currently there is there is no known optimal way of setting the parameters of these controllers. In this work, we have applied an iterative feedback tuning (IFT) method to set proportional-integral controller parameters to values that minimize a specific performance metric. The metric used is based on the residual local field potential (LFP) beta power and the stimulation intensity, rewarding maximum beta suppression with minimal stimulation intensity. We have tested this method in a computational model of parkinsonian basal ganglia, capable of modelling the pathological beta activity and simulating the LFP. We have shown that the controller parameters are updated in accordance with the predefined goals and that the behaviour of the controller is dependent on the relative importance ascribed to the beta power and the stimulation intensity.