Model Identification and Adaptive Control of Dynamic Neural systems
Howard Jay Chizeck
National Science Foundation (Center for Sensorimotor Neural Engineering)
09/15/2015 - 09/30/2017
In this project we seek to develop and test algorithms capable of modeling and co-adaptive control of dynamic neural systems. Building on the Medtronic-funded Activa PC+S/Nexus-D developments in our lab, we have a platform for chronic neural recording and stimulation at a number of sites on the cortical strip as well as the thalamic deep brain stimulation (DBS) electrode. This enables us to investigate the use of system identification methods for characterizing complex neural systems that may be subject to neural plasticity. Therefore, this work will support all three testbeds. Primarily we will support the cortical co-adaptation with BBCI testbed via development of co-adaptive closed-loop control algorithms based on identified models of cortical and thalamic function. We also support the cortico-spinal reanimation testbed, as projects in that testbed will likely make use of the same device platform with chronic electrocorticography (ECoG), using the cortical strip electrodes, as well as stimulation of spinal electrodes. These projects may benefit from identified models of the cortico-spinal pathway for operation that will be enabled by our project. Additionally, we will support the cortical and spinal plasticity testbed by providing methodologies for monitoring changes in neural systems, based upon chronic recording and simulation. In this work, we will test our system identification and co-adaptive control algorithms on a select population of Parkinson’s disease (PD) and essential tremor (ET) patients treated with DBS, as the implantation of these devices in this class of patients is FDA approved and covered by our IDE or those of our collaborators. This work could profoundly impact current DBS treatment regimens by allowing for closed-loop control and optimized stimulation protocols.