26th Annual Student Research Conference
A Novel Statistical Approach to Estimating Parameter Uncertainties in Disease Transmission Models
Colin Teberg
Dr. Jon Beck, Faculty Mentor
In 2000 the WHO launched the Global Programme to eradicate Lymphatic Filariasis (LF) in 27 countries. Eradicating LF requires a thorough understanding of the uncertainties in estimates of transmission parameters determining disease prevalence. Modern dynamic models require upwards of 20 parameters to infer disease behavior; this presents difficulties when many of these parameters cannot be measured. LF is described by robust, yet complicated models requiring more data than is feasible to gather. Parameter uncertainties are estimated from Bayesian data-model assimilation approaches. While the Bayesian algorithms are fairly accurate and robust in some instances the Bayesian methodology fails to completely estimate the uncertainties in the model parameters. Given that the re-sampling step does so with replacement incomplete uncertainty estimates yield a low unique-resample-size (URS) rendering the parameter estimates incomplete and uninformative. Here we present a novel solution to the problem of increasing URS.
Keywords: Mathematical Biology, Bayesian Melding, Lymphatic Filariasis, Epidemiology, Modelling, Capstone
Topic(s):Computer Science
Mathematical Biology
Statistics
Presentation Type: Oral Paper
Session: 402-3
Location: VH 1010
Time: 3:00