Adaptive Parameter Modulation of Deep Brain Stimulation Based on Improved Supervisory Algorithm

Abstract

Clinically deployed deep brain stimulation (DBS) for the treatment of Parkinson’s disease operates in an open loop with fixed stimulation parameters, and this may result in high energy consumption and suboptimal therapy. The objective of this manuscript is to establish, through simulation in a computational model, a closed-loop control system that can automatically adjust the stimulation parameters to recover normal activity in model neurons. Exaggerated beta band activity is recognized as a hallmark of Parkinson’s disease and beta band activity in model neurons of the globus pallidus internus (GPi) was used as the feedback signal to control DBS of the GPi. Traditional proportional controller and proportional-integral controller were not effective in eliminating the error between the target level of beta power and the beta power under Parkinsonian conditions. To overcome the difficulties in tuning the controller parameters and improve tracking performance in the case of changes in the plant, a supervisory control algorithm was implemented by introducing a Radial Basis Function (RBF) network to build the inverse model of the plant. Simulation results show the successful tracking of target beta power in the presence of changes in Parkinsonian state as well as during dynamic changes in the target level of beta power. Our computational study suggests the feasibility of the RBF network-driven supervisory control algorithm for real-time modulation of DBS parameters for the treatment of Parkinson’s disease.

Predictive neuromodulation of cingulo-frontal neural dynamics in major depressive disorder using a brain-computer interface system: A simulation study

“…Over the past decade, there has been much work in developing model-based closed-loop neuromodulation methods for neurological and neuropsychiatric disorders (Santaniello et al, 2010 ; Liu et al, 2011 ; Ehrens et al, 2015 ; Nagaraj et al, 2017 ; Bolus et al, 2018 , 2021 ; Yang et al, 2018a , 2021b ; Su et al, 2019 ; Fang and Yang, 2021 , 2022 ; Zhu et al, 2021 ). These studies share a typical framework of first conducting offline system identification to fit a model, then designing feedback controllers based on the fitted model, and finally using the controller for online closed-loop neuromodulation.…”

“…We also adopted the same framework in this work. However, prior studies have mainly focused on neurological disorders such as Parkinson's disease (PD) (Santaniello et al, 2010 ; Liu et al, 2011 ; Su et al, 2019 ; Zhu et al, 2021 ) and epilepsy (Ehrens et al, 2015 ; Nagaraj et al, 2017 ) but not neuropsychiatric disorders such as MDD. This is likely because a good understanding of the disease mechanism of MDD is still lacking (Mayberg, 1997 ; Drevets, 2001 ; Williams, 2017 ).…”

“…Recent clinical studies of mood symptoms have suggested that neural activity from multiple sites and multiple frequency bands can be related to mood (Kirkby et al, 2018 ; Rao et al, 2018 ; Sani et al, 2018 ; Smart et al, 2018 ; Scangos et al, 2021a ; Bijanzadeh et al, 2022 ; Smith et al, 2022 ; Xiao et al, 2023 ). By contrast, model-based DBS methods for PD and epilepsy usually assume that the neural activity comes from a local brain region—e.g., basal ganglia in PD (Santaniello et al, 2010 ; Liu et al, 2011 ; Su et al, 2019 ; Zhu et al, 2021 )—or a single frequency band—e.g., β band in PD (Su et al, 2019 ; Zhu et al, 2021 ). Our prior work proposes an extension to multi-input multi-output regulation of brain states (Yang et al, 2018a , 2021b ) but was tested in a general dynamic model that is not tightly connected with MDD.…”

“…More advanced predictive neuromodulation methods have been proposed and tested in simulations to address dynamic neural responses in the context of Parkinson's disease (Santaniello et al, 2010 ; Liu et al, 2011 ; Su et al, 2019 ; Zhu et al, 2021 ) and epilepsy (Ehrens et al, 2015 ; Nagaraj et al, 2017 ). Such methods first use offline system identification to build a dynamic model that quantifies the stimulation effect on neural activity and then use the model to design an online feedback controller such as a proportional-integral (PI) controller to regulate neural activity (Su et al, 2019 ; Zhu et al, 2021 ). However, these methods largely regulate neural activity at a single brain site or frequency band (see Section 4).…”

Suppression of seizure in childhood absence epilepsy using robust control of deep brain stimulation: a simulation study

“…A linear controlled auto-regressive model was employed to represent the relationship between stimulation frequency and beta band power and then was coupled with Routh-Hurwitz stability analysis to tune the coefficients of the PI controller. In 63 , a Radial basis function neural network (RBFNN) was used in a supervisory control algorithm as an inverse model of the computational model of PD to track the desired dynamic beta power. To track the model uncertainty and provide robustness to noise and disturbance in control of brain states using DBS, Fang et al 64 proposed an adaptive robust controller to cancel uncertainties included in a state-space brain network model.…”