Department of Biostatistics Seminar/Workshop Series

Estimating Bayesian Epidemiological Models using PyMC

Christopher Fonnesbeck, PhD
Department of Mathematics and Statistics
University of Otago
Dunedin, New Zealand

Wednesday, February 3, 1:30-2:45pm, MRBIII Conference Room 1220

Intended Audience: Persons interested in applied statistics, statistical theory, epidemiology, health services research, clinical trials methodology, statistical computing, statistical graphics, R users or potential users.

The Markov chain model is a natural choice for characterizing disease dynamics based on observations, such as survey data. From such models it is easy to calculate various metrics of interest, including probabilities of infection or recovery, duration of illness, expected time to infection, and life expectancy. I will present the core principles of Markov chain models, including how to implement them in a Bayesian context. Using Bayesian methods for estimation easily allows investigators to add nuance and realism to a basic MC model, without having to derive closed-form estimators for parameters of complex models. I will also introduce PyMC, a general programming toolkit for developing object-oriented Markov chain Monte Carlo models in the Python programming language, using simulated disease data as an example.