PEARL: Probing Entity Aggregation in Real Life

摘要

The classical approach for predicting tubercles bacillus (TB) prevalence falls in the partial differential equation (PDE) framework, which is essentially equal to the assumption of uniform random interaction network among people. The assumption, however, conflicts with common knowledge that some people interacts with many partners while others interact with limited partners. To accurately capture the interaction patterns among people, a mobile-based system called PEARL is proposed in this study. PEARL utilizes the characteristic that Bluetooth device has an effective range of ~10 meters, which is also the infectious distance of TB. Experimental results on several volunteers suggest that the interaction pattern roughly conforms to scale-free distribution, which helps improve prediction of prevalence of TB in China.