Design of the feedback controller for deep brain stimulation of the parkinsonian state based on the system identification

Abstract

A novel closed-loop control strategy of deep brain stimulation is explored in this paper. By establishing an input-output model of the basal ganglia, the causality between the external stimuli and neuronal activities can be revealed. One-step ahead prediction constructs the probable future information of the tracking errors, which is used to guide the amplitude of the current pulse train stimuli. By comparing the traditional and iterative learning proportional control algorithms, the latter control strategy not only automatically can optimize the control signals without requirements of any particular knowledge on the details of model, but also can reduce the energy expenditure of the stimuli by accelerating the control process. This work may point to the potential value of model-based design of closed-loop controllers and pave the way towards the optimization of deep brain stimulation parameters and structures for Parkinson's disease.