A Neural Network-Based Sensor Fault-Tolerant Smooth Super-Twisting SMC for Parkinson’s Disease Control

Deep brain stimulation (DBS) therapy is a highly effective method for controlling the symptoms of Parkinson’s disease (PD), specifically the tremors. While open-loop stimulation of a single region of the basal ganglia (BG) is considered acceptable, studies indicate that closed-loop stimulation of two regions of the BG yields higher performance. This approach not only enhances symptom control but also reduces the stimulation signal received by the brain in a closed-loop stimulation approach, thereby minimizing side effects. Therefore, this article presents the concurrent stimulation of two regions by a two-part feedback controller, in which a partial state feedback (PSF) controller stimulates the globus pallidus internal (GPi) and a radial basis function neural network (RBFNN) neural network-based smooth super-twisting sliding mode controller (NNSSTSM) stimulates the subthalamic nucleus (STN). The proposed controller effectively regulates tremors, which are a symptom of PD, even in the presence of parameter uncertainties, as well as the faults of the sensor attached to the patient’s finger to measure the tremor. This research applies two neural networks: the first is applied in an observer for estimating sensor faults, and the second is used in the controller to estimate nonlinear terms and uncertainties accurately. The stability analysis of the closed-loop system with the proposed controller and observer is conducted by the Lyapunov method. Simulation results demonstrate that the proposed controllers deliver highly effective stimulation signals to the brain with high performance, which effectively decreases the intensity of the tremors. Overall, our findings highlight the advantages of dual-region DBS with advanced control strategies, showing promise for improved PD symptom management.