Platform for Model-Based Design and Testing for Deep Brain Stimulation

Abstract

Deep Brain Stimulation (DBS) is effective at alleviating symptoms of neurological disorders such as Parkinson's disease. Yet, despite its safety-critical nature, there does not exist a platform for integrated design and testing of new algorithms or devices. Consequently, we introduce a model-based design framework for DBS controllers based on a physiologically relevant basal-ganglia model (BGM) that we capture as a network of nonlinear hybrid automata, synchronized via neural activation events. The BGM is parametrized by the number of neurons used to model each of the BG regions, which supports tradeoffs between fidelity and complexity of the model. Our hybrid-automata representation is exploited for design of software (Simulink) and hardware (FPGA) BGM platforms, with the latter enabling real-time model simulation and device testing. We demonstrate that the BGM platform is capable of generating physiologically relevant responses to DBS, and validate the BGM using a set of requirements obtained from existing work. We present the use of our framework for design and test of DBS controllers with varying levels of adaptation/feedback. Our evaluations are based on Quality-of-Control metrics that we introduce for runtime monitoring of DBS effectiveness.