The overall goal of the project is to complement the principal investigator's epidemiologic analytical skills by providing an environment for continued development as a veterinary epidemiologist capable of addressing complex analytical issues competently and confidently.
The specific aims are: 1) Broaden my knowledge of analytical epidemiologic techniques, namely, the application of random effect models to observations (including serial observations) with binary outcomes. The principal investigator will travel to Maisons Alfort to interact with a group of scientists who have a vast experience with the application of random effect models to veterinary data analysis. 2) Working with Dr. Sanaa and his colleagues, assess the impact of the presence of intra-group correlation on risk identification among observations collected in distinct veterinary epidemiologic studies involving herds of cattle and horses in stables. All these studies investigate etiologic agents that are known to pose hazards to human health: Cryptospordium, Giardia, and Leptospira. The intra-group correlation (herds of cows, horse in a stable) will be estimated from the binary data random effect models. 3) Evaluate the strengths and the limitations of the three currently developed binary data random effect (BDRE) models in risk analysis. These BDRE are the logistic-binomial, logistic-normal, and the probit normal. The goodness of fit for each type of model when applied to each if these data will be compared. The characteristics of a data set that suggest the usage of one model over another will be identified. This evaluation will be carried out on the data that have already been collected in the US and data available in France. 4) Learn the approaches to the analysis of survival data including the use of fraility models.
