A Bayesian Maximum Entropy Framework for Real-Time Infectious Disease Prediction

• Activity Number: 646
• Type: Topic Contributed
• Date/Time: Thursday, August 2, 2012 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics and the Environment
• Author(s): Hwa-Lung Yu*+ and George Christakos and Jose Angulo
• Companies: National Taiwan University and San Diego State University and University of Granada
• Keywords: Bayesian maximum entropy ; data assimilation ; Kalman filter ; SIR model

Abstract

Emerging infectious diseases, such as SARS, H1N1, and H5N1, have been a major public health concern in this century. Understanding the spatiotemporal patterns of the disease spreading plays an important role in disease prevention and control. Mathematical models have been developed for the modeling of spatiotemporal spreading of infectious epidemics, among which the susceptible-infected-recovered (SIR) model is one of the most popular models. For the cases of emerging infectious diseases, they are usually characterized by the uncertain number of disease occurrences and the unknown parameters for most of the mathematical models while the diseases are occurring. However, these information are very important to the control of space-time disease spreading. This study proposes a Bayesian maximum entropy framework integrate the physical-meaning infectious model, i.e. SIR model, and the uncertain number of the disease occurrences during the epidemics. The proposed framework can efficiently update the unknown parameters and perform the spatiotemporal prediction of infectious diseases in a real-time manner.