NEC Seminar: Friday, 12/5/2025, 9:00 AM
Speaker: Madelaine Blincoe
Advisor: Dr. Ronald Triolo
Title: Decoding User Intent to Modulate Walking Speed After SCI
Abstract: Peripheral nerve stimulation is a powerful tool for improving ambulation after spinal cord injury (SCI). Yet, current systems to facilitate gait rely on feedforward control of stimulation that enforce one clinically optimized walking speed for each user. This lack of adaptability hinders safety and community ambulation, where modulating speed is essential to avoid obstacles and maximize social engagement. To address this issue and develop new control systems that enhance the ability of individuals with SCI to intuitively and automatically vary their gait speeds continuously or by switching between patterns of stimulation based on the kinematics of the Center of Mass (CoM), we first investigated the ability to modulate speed through voluntary interactions with an assistive device during pre-programmed open-loop stimulation optimized for a preferred step rate (cadence). Preliminary experiments confirmed that participants with SCI are able to modulate walking speed by voluntarily adjusting step-length by pushing or pulling on the walker. Voluntary speed changes were reflected in signals from VICON motion capture CoM velocity and wearable inertial measurement units (IMUs), and were considered sufficient to be tested as a command input to a control system that would adjust timing and intensity of stimulation to best achieve the desired walking speed and reduce excessive forces being applied to the walker. We then applied a First-Order Hidden Markov Model to determine the probability of a user exhibiting and transitioning between three discrete intended gait speeds from the CoM velocity and IMU signals, and evaluated classification accuracy and predictive latency. Preliminary results indicate the velocity of the CoM and acceleration of the sacrum were the most accurate biomarkers of intended gait speed, achieving prediction accuracies exceeding 90% with a latency of less than one step period, which would be appropriate for switching between stimulation patterns clinically optimized for each walking speed. This presentation focuses on the development of this controller framework and its critical role in driving a future real-time controller that modulates stimulation pulse width, amplitude, and timing based on user intent to change walking speed.