Identification, development, and assessment of prosthesis behaviors to improve the stumble recovery outcomes of transfemoral prosthesis users

Abstract

While the functionality and safety of lower limb prostheses has improved in recent decades, a notable deficit remains when recovering from a trip or stumble. Transfemoral prosthesis users are at a substantially heightened risk of falling compared to healthy counterparts. This is in part due to the fundamental function of currently commercially available prostheses, most of which are low impedance, passive devices. The knee’s low impedance allows the knee joint to be driven by the residual limb, giving the user direct control over the prosthesis, while the passive system reduces cost, weight, and control complexity. However, this also means that a perturbation during swing phase (e.g., a stumble) can easily disturb the intended trajectory of the prosthesis, and with no powered elements to control the knee joint, the prosthesis has little to no ability to correct the disturbed trajectory. Through this work, the stumble recovery response of transfemoral prosthesis users with commercially-available prostheses is characterized, along with its deficits, to inform a powered, mechatronic intervention to reduce fall risk and improve the quality of recovery. To develop the intervention, first a laboratory-based, swing phase stumble perturbation system is created to enable precise, repeated exposure to stumble perturbations across swing phase. The stumble perturbation system is validated by comparing the responses of healthy individuals to previous works, while creating a high-resolution characterization of the healthy response across swing phase. Next, the stumble recovery response of transfemoral prosthesis users with commercially-available prostheses is characterized and specific functional deficits are highlighted. The functional deficits are used to inform the design of a stumble recovery controller for a powered knee prosthesis. Two control strategies are compared, one which utilizes one strategy every time inspired by the response of current, commercially available devices, and another which utilizes two different recovery strategies inspired by healthy individuals by observing the post-perturbation impact dynamics to determine which is the best fit. Comparing the stumble recovery outcomes between the two controllers and to the commercially-available prostheses demonstrates a consistent benefit while using the bimodal stumble recovery controller, improving both the user’s center of mass control and reducing compensatory behavior. Overall, the mechatronic intervention shows promise in its ability to improve stumble recovery outcomes in the transfemoral prosthesis user population.