Closing the loop in DBS: A data-driven approach.

Abstract

Deep brain stimulation (DBS) has transformed the treatment of movement disorders like Parkinson's Disease (PD). Innovations in DBS technology and experimentation have fostered adaptive DBS (aDBS), which employs a closed-loop system that senses physiological biomarkers to inform precise neuromodulation and personalized therapy. This review analyzes several promising advances in aDBS, including biomarker detection, control policies, mechanisms of efficacy, and a data-driven approach using artificial intelligence to decode motor states from neural signals. Investigations into data-driven approaches have expanded biomarker detection beyond subcortical beta oscillations, leveraging other neural and kinematic signals. Future aDBS systems that accommodate multi-modal inputs have the potential to bolster therapeutic efficacy and address symptoms not addressed by beta-driven aDBS. Continuing investigation is necessary to address existing technical and computational challenges for further clinical translation.