This dissertation includes three projects that cover two different topics; (1) The applications and comparisons of methods in statistical causal inference; (2) Modeling the spatio-temporal dependence in disease spread to improve the efficiency in regional surveillance. In the first project, we introduced and evaluated the performance of methods including inverse probability weighting (IPW), inverse conditional probability weighting (ICPW), and doubly robust in the context of an observational study of Bovine Respiratory Disease (BRD). In the second project, we applied IPW approach to National Health Interview Survey (NHIS) data from 2017, where the purpose is to obtain a less biased estimation of the association of physical activity on cardiovascular disease. To improve the accuracy of estimation for IPW approach, we also adjusted more potential confounding variables than what have been commonly adjusted for in the past literature. In the third project, we proposed a Bayesian spatio-temporal model with missing values to study the Porcine epidemic diarrhea virus (PEDV) spread across the state of Iowa. The proposed model can be used for developing better sampling guideline in detecting the incursion of new pathogens regionally.