Modeling Temporal Dynamics of Vaginal Bacterial Communities in Healthy Women

• Activity Number: 424
• Type: Contributed
• Date/Time: Tuesday, August 2, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Zaid Abdo*+ and Ursel Schutte
• Companies: University of Idaho and Indiana University
• Keywords: Hierarchical models ; Bayesian Statistics ; Mixture models

Abstract

The beneficial effects of the endogenous microbiota on women's health are numerous and poorly characterized. Significant alterations or disruptions of the vaginal microbiome, such as in bacterial vaginosis, may increase women's risk including preterm labor and acquisition of infections. We developed a hierarchical Bayesian framework in an effort to understand factors associated with maintenance and fluctuation of the vaginal microbiota over time. Our framework includes modeling changes in the microbial community structure using a Bernoulli distribution to describe presence and absence of each phylotype and a Poisson distribution to describe their abundances. Metadata--environmental, behavioral, and life history data-were linked using a generalized linear model. This approach allowed for detecting dependencies between bacterial community structures and observed metadata. Preliminary results suggest that the temporal dynamics of bacterial communities are personalized, that is there are no obvious patterns in changes in vaginal bacterial communities over time among woman but rather the dynamics are individual to each woman.