Quantification longue durée de la marche en ambulatoire : application à la maladie de Parkinson et étude exploratoire sur le syndrome post-commotionnel

Gait assessment using wearable inertial devices—combining motion sensors and analysis algorithms—represents a promising approach for objectively quantifying gait disorders in neurology. Initially developed for use in controlled environments, this methodology has progressively been extended to realworld recordings, providing a more natural and representative evaluation of patients’ gait behavior, free from the “white coat” and Hawthorne effects. This thesis fits into this context with two application cases : Parkinson’s disease, a neurodegenerative condition with welldocumented stereotyped gait impairments, and post-concussion syndrome (PCS), following a mild traumatic brain injury, explored here in a more prospective manner. A systematic review of studies on realworld gait assessment in patients with Parkinson’s disease was first conducted, followed by a clinical study involving onemonth recordings using a wearable inertial device worn daily by three groups : Parkinson’s patients, PCS patients, and healthy controls. The cross-sectional analysis of 17 Parkinson’s patients, 4 PCS patients, and 5 control participants demonstrates good device acceptability among Parkinson’s patients. In this group, the digital mobility outcome SV95C (95th percentile of stride velocity), computed from ambulatory data, showed a strong correlation with standard clinical gait tests (r = 0.72 with the 6- Minute Walk Test ; r = 0.65 with the 10- Meter Walk Test ; p