This thesis aims to identify the spatio-temporal patterns of Tuberculosis prevalence at county level in Jiangsu province, China during 2011--2021. We established Bayesian hierarchical Poisson models with mixed effects, combining fixed effects associated with demographic factors, medical conditions, socio-economic factors, and SPDE-AR(1) spatio-temporal random effects. The optimal models are selected based on DIC, WAIC and correlation coefficient using integrated nested Laplace approximation based algorithm, a faster and more accurate alternative of MCMC. The key factors influencing TB prevalence and hot spots are identified. The findings may help the TB risk management in the allocation of medical sources and strategies.