Analyse des mouvements 3D en temps réel pour un dispositif médical destiné au maintien de l'indépendance fonctionnelle des personnes âgées à domicile

摘要

We propose in this manuscript a realtime3D movement analysis system for inhomefunctionalabilities assessment in aged adults. As a first step, the purpose is to maintain the functionalindependence of this population and to allow an earlier detection of a motor decompensation inorder to facilitate a rehabilitation process. To quantify the equilibrium quality of a subject, webuilt a system using the Kinect sensor in order to analyze a simple clinical test validated in geriatricrehabilitation: the Timed Up and Go (TUG). Three experiments conducted in heterogeneousenvironments (laboratory, day hospital and home) showed good measurement reliability of theidentified parameters. In particular, they allow to assign a motor control note indicating themotor frailty. Then, we proposed a video processing chain to increase the robustness of theanalysis of the various TUG phases: automatic detection of the sitting posture, patientsegmentation and three body joints extraction. The results of this work allow us to considerseveral perspectives. First, we believe conduct experiments on a larger population in order toconfirm the system reliability. Then, various technical and ergonomic improvements would benecessary to facilitate general public use. Finally, it would be interesting to extend the proposedmethodology for other clinical test to prolong the autonomy at home.