PREDICTING ABNORMAL MICROBIAL-POPULATION LEVELS IN THE VAGINAL ECOSYSTEM

We have attempted to predict the occurrence of abnormal microbial conc entrations in the vaginal ecosystem by fitting a statistical model to an in vivo data set containing qualitative and quantitative informatio n from women with known normal and abnormal microbial population level s. Predictive factors were selected by multiple regression with backwa rds elimination and included pH and concentrations of total aerobic ba cteria, Staphylococcus spp., Lactobacillus spp. and Prevotella spp. Th e resulting model calculates the likelihood of an abnormal microbiota, using these factors for a given individual. Potentially, this model c an be employed for clinical studies in which determining the effect of a particular alteration of the vaginal environment is important.