Fuzzy ambient intelligence in home telecare

Abstract

Telecare is the use of communication and/or sensor technologies to detect remotely the requirements of people in need of medical care or other assistance. Typically, but not exclusively, the users of telecare systems are elderly people who would otherwise need residential or nursing support. There is growing interest in the use of telecare, particularly in countries where facing growth in the proportion of elderly people in the population (with consequent increases in care requirements). Both socially and financially, it is generally preferable for the elderly to remain in their own homes for as long as possible. A number of research projects have looked into home-based telecare and telemedicine systems as a way of increasing quality of life for the elderly as well as reducing the cost of care. We can distinguish telemedicine as a subfield of telecare, where the specific aim is to remotely monitor physiological parameters of a person (such as blood sugar levels, blood pressure, etc) whereas telecare is a less specific form of monitoring looking to generate alerts in emergency situations. Telecare is not a new idea - simple alarms operated by pull-cords or pendants have been available for 30 or more years. We refer to these as first generation systems, typically used as panic-alarms to summon help in the case of a fall or other emergency. Whilst such systems have obvious benefits, they become useless when the user is unable to raise the alarm (e.g. because of unconsciousness) or does not recognise the need to signal an alarm. Second generation telecare systems make use of sensors to detect emergency situations. These sensors can be worn on the body, measuring factors such as respiration, pulse, etc, or can be sited around the home, detecting movement, possible falls, etc. Second generation systems are obviously far more sophisticated than the first generation and may require substantial computation and a degree of ambient intelligence to establish when an emergency situation arises. As with the first generation, an alarm can be triggered to summon help. In both first and second generation systems, the aim is to detect an emergency and react to it as quickly as possible. Third generation telecare systems adppt a more pro-active approach, giving early warning of possible emergency situations. In order to do this, it is necessary to monitor the well-being of a person - defined in terms of their physical, mental, social and environmental status. By detecting changes in the daily activities of the person, it is possible to detect changes in their well-being which may not be immediately observable, but can be detected over a longer period of time. Advances in sensor design and the continuing increase in processing power make it possible to implement such an ambient intelligence system, and we will describe the implementation of a third generation telecare system which has been tested in homes of elderly clients over a long term (6 to 18 months). The system is installed within a home as a customised sensor network, able to detect a persons movements and their use of furniture and household items. The sensors are designed to operate discreetly, such that the occupant need not interact directly with any component. We will focus particularly on the intelligent processing which enables the system to take low level data (e.g. kitchen sensor activated; cold water run for 20 seconds; kettle switched on for 60 seconds; fridge door opened) and answer questions such as is the occupant eating regularly, is the occupants social interaction increasing or decreasing, has the occupants sleep pattern changed significantly in the past few months etc. This is a very substantial inference and learning task and the nature of the data and queries make a soft computing approach a natural choice. Initial trials of the system indicate that intelligent data analysis and reasoning enables us to make plausible inferences of this type with a high degree of accuracy. © Springer-Verlag Berlin Heidelberg 2006.