Predicting end-point locomotion from neuromuscular activities of people with spina bifida: a self-organizing and adaptive technique for future implantable and non-invasive neural prostheses.

Neural prosthesis is a promising technique to enable paralyzed patients with conditions, such as spinal cord injury or spina bifida (SB), to control their limbs independently. However, it remains unknown whether muscle activity detected from paralyzed patients can be used to predict and reproduce their altered gait patterns that can be employed to provide closed-loop feedback for neural prostheses. In this study, we recorded muscle activity of people with SB during overground walking and developed a Self-Organizing Adaptive Prediction (SOAP) technique for neural prostheses. This technique can provide 80% more accurate prediction of end-point impaired locomotion for people with SB compared to traditional robust regression. Our results suggest that control of complex neural prostheses during locomotion can be achieved by engaging muscle activity as intrinsic feedback to generate end-point leg movement.