After a neuromotor injury, such as a stroke or spinal cord injury, patients are faced with decreased independence and quality of life due to limited hand function. Functional electrical stimulation (FES) therapy can improve patients' recovery. However, this technology is underutilized in clinics due to complex and long setup times, and there is a distinct lack of commercially available FES systems for hand therapy. In this work, we present a closed-loop, bidirectional FES system specifically designed for grasping rehabilitation. The proposed FES system leverages a first-order, adapting local model of muscle activation dynamics to control grip force by modulating the FES amplitude. We demonstrate with 12 healthy participants that the adaptive controller provides more accurate control than an autotuned PID controller. We also validate that the FES system works with a C5 spinal cord injury (SCI) participant exhibiting a typical upper motor neuron pattern of spasticity. Each participant also completed a longer-duration fatiguing experiment, and we observed that not only does the controller adapt to accurately control contractions as the activated muscles fatigue, but the model parameters are strongly correlated with fatigue and could be used to measure fatigue in real time. Finally, we show that a previously untrained user can set up the FES system in under 5 minutes. These results demonstrate the benefits of using adaptive models for FES control and can be used as a guide to design effective, translatable FES systems.
Trout et al. (Thu,) studied this question.