DescriptionThis thesis explores novel solutions that combine systems science and computational data science to improve modeling outcomes in environmental health applications. Prototype hybrid frameworks are developed to address important issues of three related topics: (1) Modeling air quality: a flexible/transferable Bayesian Ensemble Machine Learning (BEML) framework is established to improve the accuracy and resolution of atmospheric chemistry-transport modeling results. (2) Modeling infectious disease: a stochastic SEIR (Susceptible, Exposed, Infectious, Recovered) framework is constructed to assess effectiveness and compliance of layered exposure interventions on COVID-19 spread. (3) Modeling exposome and health: a Socioexposome-Wide Association Study (SWAS) framework is developed to characterize associations between COVID-19 health outcomes and multiple social and environmental factors. Comprehensive analysis is performed for each of these frameworks regarding their accuracy, interpretability, robustness, consistency, scales, etc. Methodologies and tools presented in the thesis will be beneficial to a wide range of applications in environmental and public health and other related fields.