A Bayesian Model of the Effect of Pseudomonas Aeruginosa and Ear Fenestration on Hearing Loss

• Activity Number: 72
• Type: Contributed
• Date/Time: Monday, August 12, 2002 : 8:30 AM to 10:20 AM
• Sponsor: Section on Bayesian Stat. Sciences*
• Author(s): Byron Gajewski*+ and Jack Sedwick and Patrick Antonelli
• Affiliation(s): St. Cloud State University and University of Florida, and University of Florida
• Keywords: Markov Chain Monte Carlo ; Rounding Error ; Relative Risk

Abstract

Selective occlusion of the semicircular canals can be used to treat inner ear disorders and to improve access to the petrous apex lesions. While currently described techniques have low rates of hearing loss, hearing loss due to injury of the semicircular canals remains a feared and poorly understood complication. The cause of hearing loss in these patients is unclear, but a suspected cause includes infection by bacteria from the middle ear. The most common pathogen in chronic middle ear infection is Pseudomonas aeruginosa (PA).

Three separate surgical experiments with various factor levels are performed on guinea pigs to help study the relationship between PA, middle ear infection, and hearing loss. These experiments are combined using a hierarchical model. Posterior integration is performed using MCMC techniques.

Following the MCMC, we perform a ranking of the factor levels, and provide posterior probabilities of significant hearing loss.

We present the details of this data analysis, which include model building, discussion of the probability for the difference in two distributions, and the building of the informative prior distributions used in the model.