A framework for pervasive and ubiquitous geriatric monitoring

Abstract

The elderly population is steadily increasing in most countries due to a decline in birth and mortality rate. The population of independently living senior citizens has also become significant. This has led to an active research focus in geriatric wellness. Apart from the usual physical and cognitive decline that proceeds with age, there are much softer aspects like dignity and ability to live independently (age-in-place). If we monitor the subject all the time by instrumenting them heavily, technology can possibly predict and prevent many abnormal situations. However, this curbs the idea of independent living. Hence, while instrumenting and monitoring is an important paradigm in Ambient Assisted Living (AAL), the need for the underlying technology to be non-intrusive or as natural as possible is also pressing. For example, passive infrared (PIR) sensors are quite effective and ubiquitous in detecting presence and movement. In this paper we investigate the opportunities in this sliding balance of intrusiveness versus wellness. We explore ubiquitous and pervasive monitoring to make useful inferences in the subject’s activities of daily living (ADL). To this end, we evaluate different type of commercially available sensors such as the Microsoft Xbox Kinect TM, video cameras and smart watches, which will augment the PIR sensors at home. Each sensor will allow the elderly to be monitored for specific wellness parameters. We propose a generic framework with algorithms developed in-house, highlighting the use of the Kinect for computing fall-risk through single limb stand (SLS) and postural stability, smart watches to detect multiple activities in an energy efficient manner using in-built sensors, and, finally, computer-vision based methods using video cameras to detect the liveliness of human beings by computing the micro motions of the body during sleep. These findings can be used for both early risk assessment and cognitive training of an individual. Our framework is also extensible to allow the integration of various sensors and algorithms in addition to being customizable for the context in question. A key challenge in building such a generic wellness care framework to support multiple sensor measurements, is handling the trade-off between cost, intrusiveness, and precision.