Prediction of gait recovery as a tool to rationalize locomotor training in spinal cord injury

Our objectives were to explore the effects of robotic-orthosis (LOKOMAT) training on walking impairment recovery in subjects with incomplete spinal cord injury (SCI), and to develop robust predictors of these recovery patterns. Twelve SCI subjects with different degrees of ankle spasticity participated in a 12-session LOKOMAT training regimen. One-hour gait training sessions were provided three times per week for four weeks. Subjects were evaluated at baseline, 1, 2 and 4 weeks after training. The 10-meter and 6-min walking tests and the Time-up-and-Go tests were used to evaluate gait speed and endurance, and functional ambulation and balance. A “growth mixture” model was used to characterize different recovery patterns of these measures. Logistic regression was further used to predict these recovery patterns based on the isometric voluntary contractions (MVC) of ankle flexors and extensors at the baseline. Our results showed that subjects were separable into two different classes of recovery based on severity of their baseline impairments; subjects with a higher walking capacity at the start of training showed significant improvement over four weeks of training. Our findings demonstrated that MVCs were able to predict recovery class membership and can potentially be used as significant predictors for therapeutic functional recovery after SCI.