Adaptive architecture for assisted living systems

Recent achievements of telemedicine and surveillance techniques open new challenges for development of assisted living systems for elderly or disabled. This paper presents an universal approach to designing of multimodal health monitoring systems with regard to a paradigm of ubiquitous and personalized medicine. The design combines advantages of intelligent reprogrammable sensors, flexibility of reconfigurable networks built on human body area or embedded in building infrastructures and automatic, person-dependent decision making based on presumptions and experience represented in artificial intelligence. Considering these key features leads to a system design suitable for majority of human surveillance purposes including home care, hospices, rehabilitation and sport training. The paper also presents a prototype system designed accordingly to proposed rules and tested in some experimental setups intended to simulate volunteers' homes. The results confirm that the system adapts to environment-specific relations, provides seamless monitoring with no limit of indoor and outdoor mobility and adapts to subject's habits in recognition of normal, suspected and dangerous events.