Towards intelligent self-care: Multi-sensor monitoring and neuro-fuzzy behavior modelling

The work reported in this paper investigates the problem of intelligent patient monitoring, whereby a combination of smart sensor technology and computational intelligence techniques is used to observe then understand the behavior of a person in a self-care context, such as a patient in his hospital room or an elderly in her home. The key objective is to automatically derive a model that encompasses salient details pertaining to the health and well-being of the patient monitored, and subsequently to apply decision support tools to diagnose any behavioral variation from the model and recommend remedial action. Encouraging results were obtained using a prototype system consisting of various positional sensors and a computer vision module for behavioral pattern recognition coupled with a novel self-organizing neuro-fuzzy system for classification and prediction purposes. The case study reported here is that of monitoring the motions of a patient lying in bed to detect or even anticipate falls.