The aging of the population implies an evolution of the healthcare systems, in order to take over dependent people whose proportion is growing. A possible solution is to develop and implement a home care technology. In this context, this PhD work consist to design and test a multisensory monitoring system: deployment of sensors in patient environment, sensory data fusion and automatic diagnostic algorithms, in order allowing patients to live in better safety conditions and helping health professionals to give better care quality. This manuscript presents, in the beginning, the problematic of elderly care and the state of knowledge and practices of activity analysis systems for elderly, through a literature review. Then, it describes the design step of monitoring system according to a functional approach. This approach presents the implemented hardware and software architecture in order to satisfy established specifications. We worked on two projects: 1) Homecare System aims to experiment and validate the good working of a remote monitoring system for elderly (or disabled) suffering from Alzheimer's disease. 2) BéA System (Bracelet Electronique d'Autonomie) is oriented to outdoor monitoring of able-bodied people but frail: Our contribution is to implement a system architecture which integrates a self-adaptive algorithm for detecting and modelling wandering by day-part. For these two projects, we present the results and make a first experience feedback. Finally, we propose a potential economic model that defines the available options that allow, today, the deployment these types of system.
