Influenza is a worldwide viral illness that in its mildest annual epidemic form affects 5-20% of the population and is responsible for millions of outpatient visits, nearly 300,000 hospitalizations and 36,000 deaths in the US alone . Children hold a special place in influenza epidemiology: they account for a large number of outpatient influenza or influenza-like illness visits, with higher morbidity and mortality than older children and adults; and they are also often the vectors to carry influenza to their families. Therefore, studying the transmission of influenza in school age populations through modeling and simulation studies provide the quantitative underpinning for understanding epidemics and guiding public health action, including school closures. However, modeling results are highly sensitive to input parameters pertaining to assumptions regarding contact patterns among students in and out of school. In addition, empirical data on social contacts are difficult to collect, especially when students leave the school premises.
To address these challenges, this project seeks to architect, design, implement, and evaluate new wireless sensor technology that can capture the social inter-contact matrix in school-age populations. In so doing, they will be able to provide the ground truth needed run the models with more realistic contact patterns than are available today. In particular, we propose a new generation of sensor tags that can measure the range to nearby tags and also capture posture and orientation information to better understand cliques and social structures. These new sensor tags must be small and inexpensive, and they must be usable in large numbers (e.g. 2,000 nodes deployed for a single day). They must quickly and accurately discover nearby tags, and they must efficiently store and deliver contact data for further analysis.
The first CUSP experiments used the Telos B nodes. The second CUSP experiments will use the Irene nodes. Neither the Telos nor the Irene nodes support node-to-node ranging, which is useful for determining how far apart people are from one another. Without this data (and using RSSI as a proxy), it is difficult to conclude with high certainty that two people are within infectious range. To address this deficiency, the third generation of nodes will have ranging capability.