Iterated Filtering and Its Application in Modeling Infectious Disease Dynamics

• Activity Number: 592
• Type: Contributed
• Date/Time: Wednesday, August 4, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Anindya Bhadra*+ and Edward Ionides and Karina Laneri and Mercedes Pascual
• Companies: University of Michigan and University of Michigan and University of Michigan and University of Michigan
• Keywords: Iterated Filtering ; State space models ; Disease modeling ; Maximum likelihood ; Sequential Monte Carlo

Abstract

Iterated Filtering is an algorithm for recursively computing the maximum likelihood estimate (MLE) of the parameters in a Markovian state space model using an underlying sequential Monte Carlo ?lter. A very advantageous feature of Iterated Filtering is we only need to be able to sample from the underlying state transition equations of the model, without requiring to evaluate the state transition density in any way. There are two basic advantages with this - (a) Typically the number of the hidden state variables is more than the number of the observed states and not requiring to evaluate the high dimensional density is advantageous and (b) For a continuous time diffusion process that is observed at discrete time points, state transition density may not be available in closed form. In this talk I will introduce the Iterated Filtering algorithm and its application in malaria modeling.