Learning Post-Stroke Gait Training Strategies by Modeling Patient-Therapist Interaction

Seyed Mostafa Rezayat Sorkhabadi,Mason Smith,Roozbeh Khodmbashi,Rachel Lopez,Melissa Raasch,Trent Maruyama,Christina Kwasnica,Wenlong Zhang

DOI: https://doi.org/10.1109/TNSRE.2023.3253795

2023-03-09

Abstract:For safe and effective robot-aided gait training, it is essential to incorporate the knowledge and expertise of physical therapists. Toward this goal, we directly learn from physical therapists' demonstrations of manual gait assistance in stroke rehabilitation. Lower-limb kinematics of patients and assistive force applied by therapists to the patient's leg are measured using a wearable sensing system which includes a custom-made force sensing array. The collected data is then used to characterize a therapist's strategies in response to unique gait behaviors found within a patient's gait. Preliminary analysis shows that knee extension and weight-shifting are the most important features that shape a therapist's assistance strategies. These key features are then integrated into a virtual impedance model to predict the therapist's assistive torque. This model benefits from a goal-directed attractor and representative features that allow intuitive characterization and estimation of a therapist's assistance strategies. The resulting model is able to accurately capture high-level therapist behaviors over the course of a full training session (r2=0.92, RMSE=0.23Nm) while still explaining some of the more nuanced behaviors contained in individual strides (r2=0.53, RMSE=0.61Nm). This work provides a new approach to control wearable robotics in the sense of directly encoding the decision-making process of physical therapists into a safe human-robot interaction framework for gait rehabilitation.