Contribution to the development and validation of wearable-sensor-based methodologies for gait assessment and rehabilitation of people with lower limb amputation

One key objective during the rehabilitation of people with lower-limb amputation fitted with a prosthesis is the restoration of a physiological and energy-efficient gait pattern minimizing falling risks due to the loss of balance. Few practical tools are available to provide quantitative data to assist the follow-up of patients in the clinical routine. The development of wearable sensors offers opportunities to quantitatively and objectively describe gait in ecological situations. In this context, the aim of the thesis is to contribute to the development of wearable tools and protocols to support the functional rehabilitation of lower-limb amputees by providing clinically relevant quantitative data. Two complementary approaches have been implemented. The first approach consists in developing biomechanical models of the human body in order to retrieve biomechanically founded parameters. A protocol allowing to accurately estimate the body center of mass acceleration and instantaneous velocity has therefore been proposed based on gait data of ten people with transfemoral amputation and was validated in one person with transfemoral amputation. The second approach consists in identifying patterns in the signals measured by wearable sensors to extract concise descriptors of gait symmetry and dynamic balance. The clinical relevance and reliability of these descriptors have been investigated for the first time in people with lower-limb amputation. The work produced in the course of this thesis has contributed to the clinical transfer of wearable sensors into the clinical practice through the identification of clinically and biomechanically relevant parameters and the validation of original algorithms allowing to quantitatively describe the gait of lower-limb amputees.