Effect of Biofeedback When Walking on an Adaptive Split-belt Treadmill

• Sophia Burrowes, Biomedical Engineering, University of Delaware

• Jill Higginson, Mechanical Engineering, University of Delaware

Up to 80% of stroke survivors have low mobility and weak muscle groups causing a slower gait speed along with lower propulsion compared to healthy individuals. Some rehabilitation techniques are split-belt treadmills (two belts where each leg walks at different speeds) and visual feedback. Previous studies show that biofeedback increases propulsion on split-belt treadmills but the combination is limited as most treadmills have fixed speeds. We have developed an adaptive split-belt treadmill (sATM) that targets unilateral propulsion. In this study, we observed the effect of biofeedback on how participants walked on an sATM. We hypothesized that biofeedback would increase propulsion.

Four young healthy participants participated in the study. Participants first performed two familiarization trials on the Tied adaptive treadmill (ATM) and then with visual feedback on the sATM. One Tied ATM and four randomized trials on the sATM followed. The speed of the Tied ATM updated at every step from user gait mechanics. The sATM belt speed updated independently based on that leg propulsion and position on the treadmill. The propulsion gain on an assigned side was varied for the sATM – Equal, Low, Medium, and High imposing an increased propulsion demand. Visual feedback of the belt speed difference was displayed with a target (bounds =  ± 0.1 m/s).

The peak AGRF for the assigned side was calculated. Trials with over 70% cross-over- when the leg contacts both belts instead of one, were excluded from analysis. For all for participants, the effect size ω2from the ANOVA was relatively small (-0.0055). Without biofeedback, the effect size ω^2 was moderate (0.22). Success scores were calculated and two participants who scored less than 50% in sATM trials were excluded from the analysis. After excluding them, the effect size ω^2for peak AGRF was high (0.72). 

There was almost zero effect with biofeedback; however excluding the last two participants showed there is a high effect on propulsion which suggests that biofeedback changes peak propulsion. Visual feedback and a split-belt treadmill may change gait mechanics which means participants’ paretic side increased propulsion. This may increase peak AGRF in clinical populations such as stroke.