Project Summary/Abstract
Neurophysiological and mental ailments such as epilepsy, Parkinson’s disease (PD), schizophrenia, and
Alzheimer’s disease affect millions of lives and cost the US economy more than 100 billion dollars yearly in lost
productivity. Deep brain stimulation (DBS) has emerged as a potential therapy for disrupting pathological
synchronous firing patterns of neurons and restoring healthy oscillations in these disorders when
pharmacological interventions fail. However, existing DBS devices require extensive clinical interventions in
tuning stimulation parameters over the treatment period. Automation of the tuning of stimulation parameters
adaptively based on the disease state and underlying stimulation-induced neural plasticity has been a long-
standing question in DBS therapy. In this project, we will address these questions in in-silico studies through
formal closed-loop model-based optimal control techniques and machine learning. The overall objective of this
project is to develop a closed-loop DBS computational framework for rigorous testing of DBS strategies and
design real-time feedback-based multi-input multi-output (MIMO) DBS techniques to provide a safe and long-
lasting effect on the desynchronization of pathological neuronal activity in PD. In Aim 1, we will develop novel
open-loop DBS techniques targeting specific forms of neural plasticity to provide long-lasting post-stimulus
desynchronization of neural activity and investigate biological mechanisms underlying DBS. In Aim 2, we will
develop a new class of MIMO closed-loop DBS techniques, using network synchrony and oscillations’ power and
phase as electrophysiological biomarkers of the disease state as feedback signals to desynchronize the excessively
(pathologically) synchronized neuronal activity and suppress abnormal disease-specific neural oscillations in
large-scale models of PD and hippocampal CA1 circuit. In Aim 3, we will develop the necessary software tools to
integrate electrophysiology hardware with the developed closed-loop DBS framework. To accomplish these
goals, we will leverage techniques from model-based closed-loop optimal control, large-scale optimization, and
machine learning. The accomplishment of these goals will bring tools from closed-loop analysis and machine
learning into the realm of neuronal desynchronization and oscillations. This will enhance our fundamental
understanding of the role of closed-loop DBS in manipulating neuronal synchrony and rhythms by harnessing
neural plasticity, ultimately enabling the development of novel neuromodulation protocols for treating a class of
brain disorders such as PD and epilepsy that are putatively caused by aberrant neural synchronization.
