Probabilistic Situation Modeling from Ambient Sensors in a Health Condition Monitoring System

Abstract

The abundance of sensors in daily life infrastructures and mobile devices can allow to determine what the users are doing, which is the situation of the environment they are in, and therefore what needs they can have and take action accordingly. Artificial Intelligence techniques are applied in order to give the users the functionality that best suits their needs. This is what is called “context-aware computing”. The term “Ambient Intelligence” refers to this technology and emphasizes the incorporation of local intelligence to computing components. Ambient Intelligence is a huge field that goes from the acquisition of data from the environment, to fusioning the gathered information and data, to extracting situation characteristics, and to finally selecting and providing adequate information and services based on the extracted context. There are many applications of this technology. In this research paper, we present a Temporal Probabilistic Graphical Model based on Context Extraction Modules for Situation Modeling applications. This model is implemented and analyzed in the context of a Health Condition Monitoring System for recognizing and keeping track of changes in the Activities of Daily Living, an elderly care indicator used to detect emerging medical conditions.