Characterization of an Algorithm for Autonomous, Closed-Loop Neuromodulation During Stroke Rehabilitation

Abstract

<h3>Research Objectives</h3> Main Objective: To characterize a novel algorithm capable of triggering Vagus Nerve Stimulation (VNS) on large movements during post-stroke physical rehabilitation. Secondary Objective: To investigate a set of parameters that optimizes triggering rate and movement selectivity for the novel algorithm. <h3>Design</h3> We sought to leverage sensors and mathematics to design a system that automatically triggers stimulation based on parameters of movement during the rehabilitation exercise. To develop the algorithm, we analyzed data collected from 12 stroke patients with handheld sensors that record compression force, translation, rotation, and touch accuracy. We applied several algorithms with unique conditional metrics and evaluated which movements the algorithms selected. We characterized the effects of parameter sweeping within the algorithms to influence the rate and selectivity of stimulation. Based on this analysis, we chose a single algorithm that effectively maximized optimal trial selection with desired timing intervals. <h3>Setting</h3> Institutional study with the general community. <h3>Participants</h3> Subjects include adult chronic stroke patients, aged 22-79, whose stroke occurred at least 12 months prior to study enrollment. <h3>Interventions</h3> Not applicable. <h3>Main Outcome Measures</h3> Optimal trial selection with desired timing intervals. <h3>Results</h3> The choice algorithm continually adapts over time to adjust for intermittent periods of rest and person-to-person variability. A variety of electronic signals from various controllers (handheld sensors, touchscreen, and keyboard) can be processed by the algorithm while maintaining similar triggering goals. <h3>Conclusions</h3> Our findings demonstrate this strategy to select optimal trials holds promise to overcome limitations related to manual VNS triggers. <h3>Author(s) Disclosures</h3> None.