Title: Statistical modeling for prospective surveillance: paradigm, approach, and methods
Speakers: Al Ozonoff and Paola Sebastiani, Boston University School of Public Health
Date: March 20, 2006 12:00 - 1:30 pm
Location: DIMACS Center, CoRE Bldg, Room 431, Rutgers University, Busch Campus, Piscataway, NJ
Traditionally, statistical modeling of influenza data has focused on "excess mortality" defined by a cyclic regression model to account for strong seasonal effects. Such methods may not be appropriate for prospective surveillance, where the goal is to determine if current influenza activity departs from what would be expected based on historical data. We first discuss the prospective surveillance paradigm and its challenges which require further modeling efforts. We then detail our approach using Hidden Markov models (HMMs). Finally, we discuss other methods for handling influenza data, and discuss implications for future work in this area of research.
see: DIMACS Computational and Mathematical Epidemiology Seminar Series 2005 - 2006