Abstract
Many disease infection cases are acquired and spread in hospital settings. Early detection of such infection cases is necessary to identify population at risk accurately and effectively. This paper introduces a systematic infection detection framework utilizing mobile communication technologies including mobile networking and encounter statistics during infection breakouts. Our target environment is smart connected hospitals using Internet of things. First, the disease infection trace-back problem is defined. Then a detailed practical framework is devised using wireless encounters to facilitate the trace-back to the infection source and identify the population at risk; the nodes that are likely infected. Trace back and filtering methods are then proposed using probabilistic forward and backward search techniques. We use extensive WLAN campus traces of six buildings with different mobility characteristics and over 34K users to drive our simulations. Metrics of true positive, true negative, accuracy and coverage are measured and presented. Our results thus far show promise, where in most cases the accuracy of correctly selecting the infected nodes that have been in the area during the epidemic period were found to reach 70-80% with population selection coverage close to the infection percentage of nodes.