Inverse problem analysis is based on accurately calculating the external loads or boundary conditions that can generate a known amount of strain, stresses or displacements at pre-determined locations on a structure. In this study a backpropagation artificial neural network (ANN) is designed and validated to predict the interfacial pressures at the residual limb/socket interface from strain data collected from the socket surface. The subject of this investigation was a 45-year-old male unilateral trans-tibial (below-knee) traumatic amputee who had been using a prosthesis for 22 years.
When comparing the ANN predicted interfacial pressure on 16 patches within the socket with actual pressures applied to the socket there is shown to be 8.7 % difference, validating the methodology. Investigation of varying axial load through the subject's prosthesis, alignment of the subject's prosthesis, and pressure at the limb/socket interface during walking demonstrates that the validated ANN is able to give an accurate full-field study of the static and dynamic interfacial pressure distribution.
To conclude, a methodology has been developed that enables a prosthetist to quantitatively analyse the distribution of pressures within the prosthetic socket in a clinical environment. This will aid in facilitating the ¡°right first time?approach to socket fitting which will benefit both the patient in terms of comfort and the prosthetist, by reducing the time and associated costs of providing a high level of socket fit.