Monitoring the suitability of the fit of a lower-limb prosthetic socket using artificial neural network in commonly encountered walking conditions.

Abstract

Prosthetic sockets are still routinely designed without the aid of quantitative measurement, relying instead on the experience and skill of clinicians. Sockets remain the most common cause for complaint regarding the suitability of a prosthesis, and poor pressure distribution is implicated in many forms of unacceptable care outcomes. Monitoring pressure distribution has been effectively restricted to laboratory settings, and only limited work has examined conditions other than flat walking. In this work, a transtibial amputee completed static and dynamic tasks on flat ground, on slopes and with changes to prosthetic materials and alignment. This was achieved using a set of wireless measurement nodes and custom LabView and MATLAB code, using external strain measurements and a neural network to understand the internal pressure distribution. Future work will focus on modifying the software to be more user-friendly for a clinical operator, and in simplifying the required hardware. Although the system in its current form facilitated the desired measurements effectively, it required engineering support to function accurately. Improving the reliability and stability of the system will be necessary before routine use is possible.