Publication Date

Spring 2026

Degree Type

Master's Project

Degree Name

Master of Science in Computer Science (MSCS)

Department

Computer Science

First Advisor

Genya Ishigaki

Second Advisor

Gautam Kumar

Third Advisor

William Andreopoulus

Keywords

Closed-loop deep brain stimulation (cl-DBS), reinforcement learning (RL), desynchronization, spiking neural network, spike-timing-dependent plasticity

Abstract

Deep Brain Stimulation (DBS) is an effective and common treatment for neurophysiological and mental disorders, particularly Parkinson's Disease (PD) and epilepsy. While existing DBS strategies focus on suppressing the pathological synchronous firing patterns of neurons associated with such ailments, a previous work introduced \textit{Forced Temporal Spike-Time Stimulation} (FTSTS), a novel neurostimulation motif that has shown remarkable promise in long-lasting desynchronization of excessively synchronized neuronal firing patterns by harnessing synaptic plasticity. The use of reinforcement learning (RL) could be an effective control strategy in closed-loop DBS systems for automatic parameter tuning, while such a control system has not been developed for clinical applications. % such a system is yet to be developed for clinical applications, and no such system has been proposed which exploits the dynamical aspects of the underlying network that create the pathological symptoms. This research instantiates an Excitatory-Inhibitory (EI) Network model as an interactive environment suitable for RL models. A Proximal Policy Optimization (PPO) RL agent is established as the closed-loop controller for choosing optimal stimulation parameters to disrupt pathological synchronous firing patterns of neurons and to reduce the overall power dissipation required by the DBS strategy. In summary, this work introduces a novel approach to implementing an adaptive parameter-tuning for a closed-loop DBS system, leveraging the advantages of FTSTS and PPO to further open the door to a novel neurostimulation therapy for epilepsy and other related medical conditions.

Download

Available for download on Sunday, May 23, 2027

Share

COinS