Enabling closed-loop DBS with RL from a BGT neural circuit simulation model for Parkinson's Disease

What is it about?

This study presents a novel real-time closed-loop DBS system utilizing a TD3 reinforcement learning agent and CBT neural circuit simulation model, achieving significant power savings while optimizing therapeutic effects for Parkinson's disease and avoiding the over-stimulation effect.

Featured Image

Why is it important?

Using 130 Hz high-frequency stimulation to STN nuclei has been the current clinical DBS treatment standard protocol. However, individual differences in symptoms and lesion sites require lengthy procedures for optimization. By leveraging RL techniques, our proposed simulation system aims to enhance therapeutic efficacy and energy efficiency, paving the way for more personalized and effective treatments in clinical settings. This research contributes to optimizing cl-DBS algorithms with identified features and pros and cons of popular RL methods. It also lays the groundwork for future clinical applications with a mechanism-based neuronal network model, promising to enhance the quality of life for individuals suffering from movement disorders in neurodegenerative diseases.