Closed-Loop Control of Epidural Electrical Stimulation in Patients with Spinal Cord Injury and Parkinson's Disease

Over the past two decades, epidural electrical stimulation (EES) has transitioned from a treatment for chronic pain to a promising paradigm for ameliorating deficits associated with neurological disorders like spinal cord injury (SCI) and Parkinson's Disease (PD). While the therapeutic effects of EES have been established, further technological developments are needed to facilitate its integration into the daily lives of the numerous affected patients. This thesis presents a comprehensive framework for the control and delivery of EES, tailored to the distinctive needs of SCI and PD patients while emphasizing user-friendliness and rapid deployment. Despite their different etiologies, PD and SCI both widely affect locomotion. Properly timed and localized stimulation of the lumbosacral spinal cord has been demonstrated to alleviate gait deficits and help SCI patients regain partially their independence. In the first part of this thesis, we focus on a framework based on wearable devices, for the synchronization of stimulation protocols to movement intentions in both classes of patients. Specifically, we propose alternative methods for accurately detecting movement intentions using inertial measurement units (IMU) in various scenarios, surpassing existing state-of-the-art approaches. These alternatives are thoroughly validated across different patient conditions, demonstrating their efficacy. Furthermore, we present the first evidence of the effectiveness of spatiotemporal EES for a patient with Parkinson's disease using the designed framework. We measure the synergistic effect of EES with traditional treatments to alleviate gait deficits. We will demonstrate a drastic decrease in freezing-of-gait (FOG). Once established a robust framework for the synchronization of EES with movement intentions, we aim to bridge the gap between laboratory-optimized technologies and their everyday use. We investigate how to compensate for sensors' misplacements and long-term gait strategy variations and how to adapt stimulation protocols to the diverse tasks that patients encounter in their daily lives. To address those challenges, we propose a novel calibration method based on the re-alignment of low dimensional manifolds extracted from inertial measurements effectively restoring detection accuracy to over 80% regardless of the initial misplacement's magnitude. We then leverage the IMU data to design a task-specific decoder able to differentiate among 4 basic activities of daily living in PD and SCI patients, we validate its effectiveness in structured and unstructured environments highlighting the effect of delivering the appropriate stimulation for alleviating gait deficits during complex tasks. This work, combined with an AI-based algorithm for spatial optimization of gait parameters, presents a potential new gold standard framework for controlling and optimizing EES as a therapeutic intervention for locomotion impairments. Finally, we explore how task-related feat