Time course of motor learning during human-in-the-loop optimization of a prosthetic foot

People with a lower-limb amputation must undergo a process of co-adaptation with a prosthesis to achieve optimal walking performance. Human-in-the-loop optimization could identify optimal prosthetic settings, while also providing insight into the process of motor learning during prosthetic tuning. The aim of the study was to investigate the time course of motor learning of people with transtibial amputation during the human-in-the-loop optimization process of a prosthetic foot, in which the stiffness and alignment were optimized to minimize metabolic cost.

Ten people with a transtibial amputation underwent an optimization protocol while walking on an instrumented treadmill with an experimental prosthetic foot with tuneable stiffness and alignment. We aimed to minimize the metabolic cost of walking by optimizing the stiffness and alignment, using an evolutionary optimization algorithm consisting of 6 generations of 6 trials. To monitor motor learning throughout the optimization process, motor learning trials with initial standard settings were repeated after each generation. Occurrence of motor learning over time was assessed by comparing metabolic cost and walking biomechanics during motor learning trials.

Metabolic cost during the motor learning trials decreased significantly (≥ 6.8 %) over time (p = 0.01). This reduction in metabolic cost was limited to the first four generations of the optimization process (i.e., 56 min).

Motor learning of people with a transtibial amputation plays a significant role during prosthetic tuning. Motor learning extended over at least 56 min in our human-in-the-loop optimization experiment. Co-adaptation of the user should therefore be taken into account during tuning of prosthetic devices.