Deep brain stimulation (DBS) of the Basal Ganglia (BG) is an effective treatment to suppress the symptoms of Parkinson’s disease (PD). Using a closed-loop scheme in DBS can not only improve its therapeutic effects but it can also reduce its energy consumption and possible side effects. In this paper, a predictive closed loop control strategy is employed to suppress the PD in real-time. A linear multi-input multi-output (MIMO) state-delayed system is considered as a simplified model of the BG neuronal network relating the stimulation signals as inputs to the beta power of local field potentials as PD biomarkers. The effect of time delay in different areas of the BG are incorporated into this model and a real-time subspace-based identification is implemented to continuously model the state of the BG neuronal network and drive the predictive control strategy. Simulation results show that the proposed MIMO subspace based predictive controller can suppress PD symptoms more effectively and with less power consumption compared to the conventional open-loop DBS and a recently proposed single-input single-output closed loop controller.

基底神经节 (BG) 的深部脑刺激 (DBS) 是抑制帕金森病 (PD) 症状的有效治疗方法。在 DBS 中使用闭环方案不仅可以提高其治疗效果，还可以降低其能耗和可能的副作用。在本文中，采用预测闭环控制策略来实时抑制局部放电。线性多输入多输出 (MIMO) 状态延迟系统被认为是 BG 神经元网络的简化模型，将刺激信号作为输入与作为 PD 生物标志物的局部场电位的 β 功率相关联。该模型将BG不同区域的时延效应纳入该模型，并实施基于实时子空间的识别，以连续模拟BG神经元网络的状态并驱动预测控制策略。仿真结果表明，与传统的开环 DBS 和最近提出的单输入单输出闭环控制器相比，所提出的基于 MIMO 子空间的预测控制器可以更有效地抑制 PD 症状，并且功耗更低。