Human activity recognition in pervasive health-care: Supporting efficient remote collaboration
Journal of Network and Computer Applications
Academic Press Ltd.
London, UK, UK
Activity inference, Collaborative environments, Human activity recognition, Pervasive Computing, Pervasive health-care, Ubiquitous computing
Technological advancements, including advancements in the medical field have drastically improved our quality of life, thus pushing life expectancy increasingly higher. This has also had the effect of increasing the number of elderly population. More than ever, health-care institutions must now care for a large number of elderly patients, which is one of the contributing factors in the rising health-care costs. Rising costs have prompted hospitals and other health-care institutions to seek various cost-cutting measures in order to remain competitive. One avenue being explored lies in the technological advancements that can make hospital working environments much more efficient. Various communication technologies, mobile computing devices, micro-embedded devices and sensors have the ability to support medical staff efficiency and improve health-care systems. In particular, one promising application of these technologies is towards deducing medical staff activities. Having this continuous knowledge about health-care staff activities can provide medical staff with crucial information of particular patients, interconnect with other supporting applications in a seamless manner (e.g. a doctor diagnosing a patient can automatically be sent the patient’s lab report from the pathologist), a clear picture of the time utilisation of doctors and nurses and also enable remote virtual collaboration between activities, thus creating a strong base for establishment of an efficient collaborative environment. In this paper, we describe our activity recognition system that in conjunction with our efficiency mechanism has the potential to cut down health-care costs by making the working environments more efficient. Initially, we outline the activity recognition process that has the ability to infer user activities based on the self-organisation of surrounding objects that user may manipulate. We then use the activity recognition information to enhance virtual collaboration in order to improve overall efficiency of tasks within a hospital environment. We have analysed a number of medical staff activities to guide our simulation setup. Our results show an accurate activity recognition process for individual users with respect to their behaviour. At the same time we support remote virtual collaboration through tasks allocation process between doctors and nurses with results showing maximum efficiency within the resource constraints.